Don't run perf tests in parallel

.github/PULL_REQUEST_TEMPLATE/release-pr.md

@@ -10,7 +10,7 @@
 <!-- List everything that should be done **before** release, any issues / setting changes / etc -->
 
 ### Checklist after release
-- [ ] Based on the merged commits write release notes and open a PR into `website` repo ([example](https://github.com/neondatabase/website/pull/219/files))
+- [ ] Based on the merged commits write release notes and open a PR into `website` repo ([example](https://github.com/neondatabase/website/pull/120/files))
 - [ ] Check [#dev-production-stream](https://neondb.slack.com/archives/C03F5SM1N02) Slack channel
 - [ ] Check [stuck projects page](https://console.neon.tech/admin/projects?sort=last_active&order=desc&stuck=true)
 - [ ] Check [recent operation failures](https://console.neon.tech/admin/operations?action=create_timeline%2Cstart_compute%2Cstop_compute%2Csuspend_compute%2Capply_config%2Cdelete_timeline%2Cdelete_tenant%2Ccreate_branch%2Ccheck_availability&sort=updated_at&order=desc&had_retries=some)

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

@@ -47,7 +47,7 @@ runs:
         else
           key=branch-$(echo ${GITHUB_REF#refs/heads/} | tr -c "[:alnum:]._-" "-")
         fi
-        echo "KEY=${key}" >> $GITHUB_OUTPUT
+        echo "::set-output name=KEY::${key}"
 
     - uses: actions/setup-java@v3
       if: ${{ inputs.action == 'generate' }}
@@ -186,7 +186,7 @@ runs:
         aws s3 cp --only-show-errors ./index.html "s3://${BUCKET}/${REPORT_PREFIX}/latest/index.html"
 
         echo "[Allure Report](${REPORT_URL})" >> ${GITHUB_STEP_SUMMARY}
-        echo "report-url=${REPORT_URL}" >> $GITHUB_OUTPUT
+        echo "::set-output name=report-url::${REPORT_URL}"
 
     - name: Release Allure lock
       if: ${{ inputs.action == 'generate' && always() }}

.github/actions/download/action.yml

@@ -34,7 +34,7 @@ runs:
         S3_KEY=$(aws s3api list-objects-v2 --bucket ${BUCKET} --prefix ${PREFIX%$GITHUB_RUN_ATTEMPT} | jq -r '.Contents[].Key' | grep ${FILENAME} | sort --version-sort | tail -1 || true)
         if [ -z "${S3_KEY}" ]; then
           if [ "${SKIP_IF_DOES_NOT_EXIST}" = "true" ]; then
-            echo 'SKIPPED=true' >> $GITHUB_OUTPUT
+            echo '::set-output name=SKIPPED::true'
             exit 0
           else
             echo 2>&1 "Neither s3://${BUCKET}/${PREFIX}/${FILENAME} nor its version from previous attempts exist"
@@ -42,7 +42,7 @@ runs:
           fi
         fi
 
-        echo 'SKIPPED=false' >> $GITHUB_OUTPUT
+        echo '::set-output name=SKIPPED::false'
 
         mkdir -p $(dirname $ARCHIVE)
         time aws s3 cp --only-show-errors s3://${BUCKET}/${S3_KEY} ${ARCHIVE}

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

@@ -41,8 +41,8 @@ runs:
             ;;
         esac
 
-        echo "api_host=${API_HOST}" >> $GITHUB_OUTPUT
-        echo "region_id=${REGION_ID}" >> $GITHUB_OUTPUT
+        echo "::set-output name=api_host::${API_HOST}"
+        echo "::set-output name=region_id::${REGION_ID}"
       env:
         ENVIRONMENT: ${{ inputs.environment }}
         REGION_ID: ${{ inputs.region_id }}
@@ -72,10 +72,10 @@ runs:
 
         dsn=$(echo $project | jq --raw-output '.roles[] | select(.name != "web_access") | .dsn')/main
         echo "::add-mask::${dsn}"
-        echo "dsn=${dsn}" >> $GITHUB_OUTPUT
+        echo "::set-output name=dsn::${dsn}"
 
         project_id=$(echo $project | jq --raw-output '.id')
-        echo "project_id=${project_id}" >> $GITHUB_OUTPUT
+        echo "::set-output name=project_id::${project_id}"
       env:
         API_KEY: ${{ inputs.api_key }}
         API_HOST: ${{ steps.parse-input.outputs.api_host }}

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

@@ -32,7 +32,7 @@ runs:
             ;;
         esac
 
-        echo "api_host=${API_HOST}" >> $GITHUB_OUTPUT
+        echo "::set-output name=api_host::${API_HOST}"
       env:
         ENVIRONMENT: ${{ inputs.environment }}
 

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

@@ -73,13 +73,6 @@ runs:
       shell: bash -euxo pipefail {0}
       run: ./scripts/pysync
 
-    - name: Download compatibility snapshot for Postgres 14
-      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
@@ -87,8 +80,6 @@ runs:
         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')
       shell: bash -euxo pipefail {0}
       run: |
         # PLATFORM will be embedded in the perf test report
@@ -163,15 +154,6 @@ runs:
           scripts/generate_and_push_perf_report.sh
         fi
 
-    - name: Upload compatibility snapshot for Postgres 14
-      if: github.ref_name == 'release'
-      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/
-        prefix: latest
-
     - name: Create Allure report
       if: always()
       uses: ./.github/actions/allure-report

.github/ansible/.gitignore

@@ -2,6 +2,3 @@ zenith_install.tar.gz
 .zenith_current_version
 neon_install.tar.gz
 .neon_current_version
-
-collections/*
-!collections/.keep

.github/ansible/ansible.cfg

@@ -3,7 +3,6 @@
 localhost_warning = False
 host_key_checking = False
 timeout = 30
-collections_paths = ./collections
 
 [ssh_connection]
 ssh_args   = -F ./ansible.ssh.cfg

.github/ansible/deploy.yaml

@@ -1,7 +1,7 @@
 - name: Upload Neon binaries
   hosts: storage
   gather_facts: False
-  remote_user: "{{ remote_user }}"
+  remote_user: admin
 
   tasks:
 
@@ -14,8 +14,7 @@
       - safekeeper
 
     - name: inform about versions
-      debug:
-        msg: "Version to deploy - {{ current_version }}"
+      debug: msg="Version to deploy - {{ current_version }}"
       tags:
       - pageserver
       - safekeeper
@@ -36,7 +35,7 @@
 - name: Deploy pageserver
   hosts: pageservers
   gather_facts: False
-  remote_user: "{{ remote_user }}"
+  remote_user: admin
 
   tasks:
 
@@ -64,29 +63,15 @@
       tags:
       - pageserver
 
-    - name: read the existing remote pageserver config
-      ansible.builtin.slurp:
-        src: /storage/pageserver/data/pageserver.toml
-      register: _remote_ps_config
-      tags:
-      - pageserver
-
-    - name: parse the existing pageserver configuration
-      ansible.builtin.set_fact:
-        _existing_ps_config: "{{ _remote_ps_config['content'] | b64decode | sivel.toiletwater.from_toml }}"
-      tags:
-      - pageserver
-
-    - name: construct the final pageserver configuration dict
-      ansible.builtin.set_fact:
-        pageserver_config: "{{ pageserver_config_stub | combine({'id': _existing_ps_config.id }) }}"
-      tags:
-      - pageserver
-
-    - name: template the pageserver config
-      template:
-        src: templates/pageserver.toml.j2
-        dest: /storage/pageserver/data/pageserver.toml
+    - name: update remote storage (s3) config
+      lineinfile:
+        path: /storage/pageserver/data/pageserver.toml
+        line: "{{ item }}"
+      loop:
+        - "[remote_storage]"
+        - "bucket_name = '{{ bucket_name }}'"
+        - "bucket_region = '{{ bucket_region }}'"
+        - "prefix_in_bucket = '{{ inventory_hostname }}'"
       become: true
       tags:
       - pageserver
@@ -124,7 +109,7 @@
 - name: Deploy safekeeper
   hosts: safekeepers
   gather_facts: False
-  remote_user: "{{ remote_user }}"
+  remote_user: admin
 
   tasks:
 

.github/ansible/neon-stress.hosts

@@ -0,0 +1,20 @@
+[pageservers]
+neon-stress-ps-1 console_region_id=1
+neon-stress-ps-2 console_region_id=1
+
+[safekeepers]
+neon-stress-sk-1 console_region_id=1
+neon-stress-sk-2 console_region_id=1
+neon-stress-sk-3 console_region_id=1
+
+[storage:children]
+pageservers
+safekeepers
+
+[storage:vars]
+env_name = neon-stress
+console_mgmt_base_url = http://neon-stress-console.local
+bucket_name           = neon-storage-ireland
+bucket_region         = eu-west-1
+etcd_endpoints        = neon-stress-etcd.local:2379
+safekeeper_enable_s3_offload = false

.github/ansible/neon-stress.hosts.yaml

@@ -1,31 +0,0 @@
-storage:
-  vars:
-    bucket_name: neon-storage-ireland
-    bucket_region: eu-west-1
-    console_mgmt_base_url: http://neon-stress-console.local
-    etcd_endpoints: neon-stress-etcd.local:2379
-    safekeeper_enable_s3_offload: 'false'
-    pageserver_config_stub:
-      pg_distrib_dir: /usr/local
-      remote_storage:
-        bucket_name: "{{ bucket_name }}"
-        bucket_region: "{{ bucket_region }}"
-        prefix_in_bucket: "{{ inventory_hostname }}"
-    safekeeper_s3_prefix: neon-stress/wal
-    hostname_suffix: ".local"
-    remote_user: admin
-  children:
-    pageservers:
-      hosts:
-        neon-stress-ps-1:
-          console_region_id: aws-eu-west-1
-        neon-stress-ps-2:
-          console_region_id: aws-eu-west-1
-    safekeepers:
-      hosts:
-        neon-stress-sk-1:
-          console_region_id: aws-eu-west-1
-        neon-stress-sk-2:
-          console_region_id: aws-eu-west-1
-        neon-stress-sk-3:
-          console_region_id: aws-eu-west-1

.github/ansible/production.hosts

@@ -0,0 +1,20 @@
+[pageservers]
+#zenith-1-ps-1 console_region_id=1
+zenith-1-ps-2 console_region_id=1
+zenith-1-ps-3 console_region_id=1
+
+[safekeepers]
+zenith-1-sk-1 console_region_id=1
+zenith-1-sk-2 console_region_id=1
+zenith-1-sk-3 console_region_id=1
+
+[storage:children]
+pageservers
+safekeepers
+
+[storage:vars]
+env_name = prod-1
+console_mgmt_base_url = http://console-release.local
+bucket_name           = zenith-storage-oregon
+bucket_region         = us-west-2
+etcd_endpoints        = zenith-1-etcd.local:2379

.github/ansible/production.hosts.yaml

@@ -1,33 +0,0 @@
----
-storage:
-  vars:
-    console_mgmt_base_url: http://console-release.local
-    bucket_name: zenith-storage-oregon
-    bucket_region: us-west-2
-    etcd_endpoints: zenith-1-etcd.local:2379
-    pageserver_config_stub:
-      pg_distrib_dir: /usr/local
-      remote_storage:
-        bucket_name: "{{ bucket_name }}"
-        bucket_region: "{{ bucket_region }}"
-        prefix_in_bucket: "{{ inventory_hostname }}"
-    safekeeper_s3_prefix: prod-1/wal
-    hostname_suffix: ".local"
-    remote_user: admin
-
-  children:
-    pageservers:
-      hosts:
-        zenith-1-ps-2:
-          console_region_id: aws-us-west-2
-        zenith-1-ps-3:
-          console_region_id: aws-us-west-2
-
-    safekeepers:
-      hosts:
-        zenith-1-sk-1:
-          console_region_id: aws-us-west-2
-        zenith-1-sk-2:
-          console_region_id: aws-us-west-2
-        zenith-1-sk-3:
-          console_region_id: aws-us-west-2

.github/ansible/scripts/init_pageserver.sh

@@ -12,19 +12,18 @@ cat <<EOF | tee /tmp/payload
   "version": 1,
   "host": "${HOST}",
   "port": 6400,
-  "region_id": "{{ console_region_id }}",
+  "region_id": {{ console_region_id }},
   "instance_id": "${INSTANCE_ID}",
   "http_host": "${HOST}",
-  "http_port": 9898,
-  "active": false
+  "http_port": 9898
 }
 EOF
 
 # check if pageserver already registered or not
-if ! curl -sf -H "Authorization: Bearer {{ CONSOLE_API_TOKEN }}" {{ console_mgmt_base_url }}/management/api/v2/pageservers/${INSTANCE_ID} -o /dev/null; then
+if ! curl -sf -X PATCH -d '{}' {{ console_mgmt_base_url }}/api/v1/pageservers/${INSTANCE_ID} -o /dev/null; then
 
     # not registered, so register it now
-    ID=$(curl -sf -X POST -H "Authorization: Bearer {{ CONSOLE_API_TOKEN }}" {{ console_mgmt_base_url }}/management/api/v2/pageservers -d@/tmp/payload | jq -r '.id')
+    ID=$(curl -sf -X POST {{ console_mgmt_base_url }}/api/v1/pageservers -d@/tmp/payload | jq -r '.ID')
 
     # init pageserver
     sudo -u pageserver /usr/local/bin/pageserver -c "id=${ID}" -c "pg_distrib_dir='/usr/local'" --init -D /storage/pageserver/data

.github/ansible/scripts/init_safekeeper.sh

@@ -14,18 +14,18 @@ cat <<EOF | tee /tmp/payload
   "host": "${HOST}",
   "port": 6500,
   "http_port": 7676,
-  "region_id": "{{ console_region_id }}",
+  "region_id": {{ console_region_id }},
   "instance_id": "${INSTANCE_ID}",
-  "availability_zone_id": "${AZ_ID}",
-  "active": false
+  "availability_zone_id": "${AZ_ID}"
 }
 EOF
 
 # check if safekeeper already registered or not
-if ! curl -sf -H "Authorization: Bearer {{ CONSOLE_API_TOKEN }}" {{ console_mgmt_base_url }}/management/api/v2/safekeepers/${INSTANCE_ID} -o /dev/null; then
+if ! curl -sf -X PATCH -d '{}' {{ console_mgmt_base_url }}/api/v1/safekeepers/${INSTANCE_ID} -o /dev/null; then
 
     # not registered, so register it now
-    ID=$(curl -sf -X POST -H "Authorization: Bearer {{ CONSOLE_API_TOKEN }}" {{ console_mgmt_base_url }}/management/api/v2/safekeepers -d@/tmp/payload | jq -r '.id')
+    ID=$(curl -sf -X POST {{ console_mgmt_base_url }}/api/v1/safekeepers -d@/tmp/payload | jq -r '.ID')
+
     # init safekeeper
     sudo -u safekeeper /usr/local/bin/safekeeper --id ${ID} --init -D /storage/safekeeper/data
 fi

.github/ansible/ssm_config

@@ -1,3 +0,0 @@
-ansible_connection: aws_ssm
-ansible_aws_ssm_bucket_name: neon-dev-bucket
-ansible_python_interpreter: /usr/bin/python3

.github/ansible/staging.hosts

@@ -0,0 +1,25 @@
+[pageservers]
+#zenith-us-stage-ps-1 console_region_id=27
+zenith-us-stage-ps-2 console_region_id=27
+zenith-us-stage-ps-3 console_region_id=27
+zenith-us-stage-ps-4 console_region_id=27
+zenith-us-stage-test-ps-1 console_region_id=28
+
+[safekeepers]
+zenith-us-stage-sk-4 console_region_id=27
+zenith-us-stage-sk-5 console_region_id=27
+zenith-us-stage-sk-6 console_region_id=27
+zenith-us-stage-test-sk-1 console_region_id=28
+zenith-us-stage-test-sk-2 console_region_id=28
+zenith-us-stage-test-sk-3 console_region_id=28
+
+[storage:children]
+pageservers
+safekeepers
+
+[storage:vars]
+env_name = us-stage
+console_mgmt_base_url = http://console-staging.local
+bucket_name           = zenith-staging-storage-us-east-1
+bucket_region         = us-east-1
+etcd_endpoints        = zenith-us-stage-etcd.local:2379

.github/ansible/staging.hosts.yaml

@@ -1,34 +0,0 @@
-storage:
-  vars:
-    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
-    pageserver_config_stub:
-      pg_distrib_dir: /usr/local
-      remote_storage:
-        bucket_name: "{{ bucket_name }}"
-        bucket_region: "{{ bucket_region }}"
-        prefix_in_bucket: "{{ inventory_hostname }}"
-    safekeeper_s3_prefix: us-stage/wal
-    hostname_suffix: ".local"
-    remote_user: admin
-
-  children:
-    pageservers:
-      hosts:
-        zenith-us-stage-ps-2:
-          console_region_id: aws-us-east-1
-        zenith-us-stage-ps-3:
-          console_region_id: aws-us-east-1
-        zenith-us-stage-ps-4:
-          console_region_id: aws-us-east-1
-
-    safekeepers:
-      hosts:
-        zenith-us-stage-sk-4:
-          console_region_id: aws-us-east-1
-        zenith-us-stage-sk-5:
-          console_region_id: aws-us-east-1
-        zenith-us-stage-sk-6:
-          console_region_id: aws-us-east-1

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

@@ -1,32 +0,0 @@
-storage:
-  vars:
-    bucket_name: neon-staging-storage-us-east-2
-    bucket_region: us-east-2
-    console_mgmt_base_url: http://console-staging.local
-    etcd_endpoints: etcd-0.us-east-2.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: us-east-2
-    console_region_id: aws-us-east-2
-
-  children:
-    pageservers:
-      hosts:
-        pageserver-0.us-east-2.aws.neon.build:
-          ansible_host: i-0c3e70929edb5d691
-
-    safekeepers:
-      hosts:
-        safekeeper-0.us-east-2.aws.neon.build:
-          ansible_host: i-027662bd552bf5db0
-        safekeeper-1.us-east-2.aws.neon.build:
-          ansible_host: i-0171efc3604a7b907
-        safekeeper-2.us-east-2.aws.neon.build:
-          ansible_host: i-0de0b03a51676a6ce

.github/ansible/systemd/pageserver.service

@@ -1,5 +1,5 @@
 [Unit]
-Description=Neon pageserver
+Description=Zenith pageserver
 After=network.target auditd.service
 
 [Service]

.github/ansible/systemd/safekeeper.service

@@ -1,12 +1,12 @@
 [Unit]
-Description=Neon safekeeper
+Description=Zenith safekeeper
 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
-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 }}"}'
+ExecStart=/usr/local/bin/safekeeper -l {{ inventory_hostname }}.local:6500 --listen-http {{ inventory_hostname }}.local:7676 -D /storage/safekeeper/data --broker-endpoints={{ etcd_endpoints }} --remote-storage='{bucket_name="{{bucket_name}}", bucket_region="{{bucket_region}}", prefix_in_bucket="{{ env_name }}/wal"}'
 ExecReload=/bin/kill -HUP $MAINPID
 KillMode=mixed
 KillSignal=SIGINT

.github/ansible/templates/pageserver.toml.j2

@@ -1 +0,0 @@
-{{ pageserver_config | sivel.toiletwater.to_toml }}

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

@@ -1,31 +0,0 @@
-# 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: "*.us-east-2.aws.neon.build"
-
-# -- Additional labels for neon-proxy pods
-podLabels:
-  zenith_service: proxy-scram
-  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: us-east-2.aws.neon.build
-
-#metrics:
-#  enabled: true
-#  serviceMonitor:
-#    enabled: true
-#    selector:
-#      release: kube-prometheus-stack

.github/workflows/benchmarking.yml

@@ -138,31 +138,23 @@ jobs:
         SLACK_BOT_TOKEN: ${{ secrets.SLACK_BOT_TOKEN }}
 
   pgbench-compare:
-    strategy:
-      fail-fast: false
-      matrix:
-        # 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 ]
-        db_size: [ 10gb ]
-        include:
-          - platform: neon-captest-new
-            db_size: 50gb
-          - platform: neon-captest-prefetch
-            db_size: 50gb
-          - platform: rds-aurora
-            db_size: 50gb
-
     env:
       TEST_PG_BENCH_DURATIONS_MATRIX: "60m"
-      TEST_PG_BENCH_SCALES_MATRIX: ${{ matrix.db_size }}
+      TEST_PG_BENCH_SCALES_MATRIX: "10gb"
       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 }}
+
+    strategy:
+      fail-fast: false
+      max-parallel: 1
+      matrix:
+        # 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 ]
 
     runs-on: dev
     container:
@@ -187,7 +179,7 @@ jobs:
         echo "${POSTGRES_DISTRIB_DIR}/v${DEFAULT_PG_VERSION}/bin" >> $GITHUB_PATH
 
     - name: Create Neon Project
-      if: contains(fromJson('["neon-captest-new", "neon-captest-prefetch"]'), matrix.platform)
+      if: matrix.platform != 'neon-captest-reuse'
       id: create-neon-project
       uses: ./.github/actions/neon-project-create
       with:
@@ -213,9 +205,11 @@ jobs:
             ;;
         esac
 
-        echo "connstr=${CONNSTR}" >> $GITHUB_OUTPUT
+        echo "::set-output name=connstr::${CONNSTR}"
 
         psql ${CONNSTR} -c "SELECT version();"
+      env:
+        PLATFORM: ${{ matrix.platform }}
 
     - name: Set database options
       if: matrix.platform == 'neon-captest-prefetch'
@@ -234,6 +228,7 @@ jobs:
         save_perf_report: ${{ env.SAVE_PERF_REPORT }}
         extra_params: -m remote_cluster --timeout 21600 -k test_pgbench_remote_init
       env:
+        PLATFORM: ${{ matrix.platform }}
         BENCHMARK_CONNSTR: ${{ steps.set-up-connstr.outputs.connstr }}
         VIP_VAP_ACCESS_TOKEN: "${{ secrets.VIP_VAP_ACCESS_TOKEN }}"
         PERF_TEST_RESULT_CONNSTR: "${{ secrets.PERF_TEST_RESULT_CONNSTR }}"
@@ -247,6 +242,7 @@ jobs:
         save_perf_report: ${{ env.SAVE_PERF_REPORT }}
         extra_params: -m remote_cluster --timeout 21600 -k test_pgbench_remote_simple_update
       env:
+        PLATFORM: ${{ matrix.platform }}
         BENCHMARK_CONNSTR: ${{ steps.set-up-connstr.outputs.connstr }}
         VIP_VAP_ACCESS_TOKEN: "${{ secrets.VIP_VAP_ACCESS_TOKEN }}"
         PERF_TEST_RESULT_CONNSTR: "${{ secrets.PERF_TEST_RESULT_CONNSTR }}"
@@ -260,6 +256,7 @@ jobs:
         save_perf_report: ${{ env.SAVE_PERF_REPORT }}
         extra_params: -m remote_cluster --timeout 21600 -k test_pgbench_remote_select_only
       env:
+        PLATFORM: ${{ matrix.platform }}
         BENCHMARK_CONNSTR: ${{ steps.set-up-connstr.outputs.connstr }}
         VIP_VAP_ACCESS_TOKEN: "${{ secrets.VIP_VAP_ACCESS_TOKEN }}"
         PERF_TEST_RESULT_CONNSTR: "${{ secrets.PERF_TEST_RESULT_CONNSTR }}"
@@ -272,7 +269,7 @@ jobs:
         build_type: ${{ env.BUILD_TYPE }}
 
     - name: Delete Neon Project
-      if: ${{ steps.create-neon-project.outputs.project_id && always() }}
+      if: ${{ matrix.platform != 'neon-captest-reuse' && always() }}
       uses: ./.github/actions/neon-project-delete
       with:
         environment: dev

.github/workflows/build_and_test.yml

@@ -35,12 +35,12 @@ jobs:
           echo ref:$GITHUB_REF_NAME
           echo rev:$(git rev-list --count HEAD)
           if [[ "$GITHUB_REF_NAME" == "main" ]]; then
-            echo "tag=$(git rev-list --count HEAD)" >> $GITHUB_OUTPUT
+            echo "::set-output name=tag::$(git rev-list --count HEAD)"
           elif [[ "$GITHUB_REF_NAME" == "release" ]]; then
-            echo "tag=release-$(git rev-list --count HEAD)" >> $GITHUB_OUTPUT
+            echo "::set-output name=tag::release-$(git rev-list --count HEAD)"
           else
             echo "GITHUB_REF_NAME (value '$GITHUB_REF_NAME') is not set to either 'main' or 'release'"
-            echo "tag=$GITHUB_RUN_ID" >> $GITHUB_OUTPUT
+            echo "::set-output name=tag::$GITHUB_RUN_ID"
           fi
         shell: bash
         id: build-tag
@@ -78,12 +78,12 @@ jobs:
 
       - name: Set pg 14 revision for caching
         id: pg_v14_rev
-        run: echo pg_rev=$(git rev-parse HEAD:vendor/postgres-v14) >> $GITHUB_OUTPUT
+        run: echo ::set-output name=pg_rev::$(git rev-parse HEAD:vendor/postgres-v14)
         shell: bash -euxo pipefail {0}
 
       - name: Set pg 15 revision for caching
         id: pg_v15_rev
-        run: echo pg_rev=$(git rev-parse HEAD:vendor/postgres-v15) >> $GITHUB_OUTPUT
+        run: echo ::set-output name=pg_rev::$(git rev-parse HEAD:vendor/postgres-v15)
         shell: bash -euxo pipefail {0}
 
       # Set some environment variables used by all the steps.
@@ -481,7 +481,6 @@ jobs:
 
   neon-image:
     runs-on: dev
-    needs: [ tag ]
     container: gcr.io/kaniko-project/executor:v1.9.0-debug
 
     steps:
@@ -495,11 +494,10 @@ jobs:
         run: echo "{\"credsStore\":\"ecr-login\"}" > /kaniko/.docker/config.json
 
       - name: Kaniko build neon
-        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}}
+        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=${{ github.sha }} --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/neon:$GITHUB_RUN_ID
 
   compute-tools-image:
     runs-on: dev
-    needs: [ tag ]
     container: gcr.io/kaniko-project/executor:v1.9.0-debug
 
     steps:
@@ -510,12 +508,11 @@ jobs:
         run: echo "{\"credsStore\":\"ecr-login\"}" > /kaniko/.docker/config.json
 
       - 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}}
+        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=${{ github.sha }} --dockerfile Dockerfile.compute-tools --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-tools:$GITHUB_RUN_ID
 
   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
@@ -530,12 +527,11 @@ jobs:
         # 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}}
+        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=${{ github.sha }} --dockerfile Dockerfile.compute-node-v14 --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node:$GITHUB_RUN_ID
 
   compute-node-image-v14:
     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
@@ -547,13 +543,12 @@ jobs:
         run: echo "{\"credsStore\":\"ecr-login\"}" > /kaniko/.docker/config.json
 
       - 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}}
+        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=${{ github.sha }} --dockerfile Dockerfile.compute-node-v14 --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v14:$GITHUB_RUN_ID
 
 
   compute-node-image-v15:
     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
@@ -565,11 +560,11 @@ jobs:
         run: echo "{\"credsStore\":\"ecr-login\"}" > /kaniko/.docker/config.json
 
       - 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}}
+        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=${{ github.sha }} --dockerfile Dockerfile.compute-node-v15 --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v15:$GITHUB_RUN_ID
 
   promote-images:
     runs-on: dev
-    needs: [ tag, neon-image, compute-node-image, compute-node-image-v14, compute-node-image-v15, compute-tools-image ]
+    needs: [ neon-image, compute-node-image, compute-node-image-v14, compute-node-image-v15, compute-tools-image ]
     if: github.event_name != 'workflow_dispatch'
     container: amazon/aws-cli
     strategy:
@@ -582,9 +577,8 @@ jobs:
 
     steps:
       - name: Promote image to latest
-        run: |
-          export MANIFEST=$(aws ecr batch-get-image --repository-name ${{ matrix.name }} --image-ids imageTag=${{needs.tag.outputs.build-tag}} --query 'images[].imageManifest' --output text)
-          aws ecr put-image --repository-name ${{ matrix.name }} --image-tag latest --image-manifest "$MANIFEST"
+        run:
+          MANIFEST=$(aws ecr batch-get-image --repository-name ${{ matrix.name }} --image-ids imageTag=$GITHUB_RUN_ID --query 'images[].imageManifest' --output text) && aws ecr put-image --repository-name ${{ matrix.name }} --image-tag latest --image-manifest "$MANIFEST"
 
   push-docker-hub:
     runs-on: dev
@@ -603,19 +597,19 @@ jobs:
           echo "{\"credsStore\":\"ecr-login\"}" > /github/home/.docker/config.json
 
       - name: Pull neon image from ECR
-        run: crane pull 369495373322.dkr.ecr.eu-central-1.amazonaws.com/neon:${{needs.tag.outputs.build-tag}} neon
+        run: crane pull 369495373322.dkr.ecr.eu-central-1.amazonaws.com/neon:latest neon
 
       - 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
+        run: crane pull 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-tools:latest 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
+        run: crane pull 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node:latest 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
+        run: crane pull 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v14:latest compute-node-v14
 
       - name: Pull compute node v15 image from ECR
-        run: crane pull 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v15:${{needs.tag.outputs.build-tag}} compute-node-v15
+        run: crane pull 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v15:latest compute-node-v15
 
       - name: Pull rust image from ECR
         run: crane pull 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned rust
@@ -625,11 +619,9 @@ jobs:
           (github.ref_name == 'main' || github.ref_name == 'release') &&
           github.event_name != 'workflow_dispatch'
         run: |
-          crane copy 369495373322.dkr.ecr.eu-central-1.amazonaws.com/neon:${{needs.tag.outputs.build-tag}} 093970136003.dkr.ecr.us-east-2.amazonaws.com/neon:latest
-          crane copy 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-tools:${{needs.tag.outputs.build-tag}} 093970136003.dkr.ecr.us-east-2.amazonaws.com/compute-tools:latest
-          crane copy 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node:${{needs.tag.outputs.build-tag}} 093970136003.dkr.ecr.us-east-2.amazonaws.com/compute-node:latest
-          crane copy 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v14:${{needs.tag.outputs.build-tag}} 093970136003.dkr.ecr.us-east-2.amazonaws.com/compute-node-v14:latest
-          crane copy 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v15:${{needs.tag.outputs.build-tag}} 093970136003.dkr.ecr.us-east-2.amazonaws.com/compute-node-v15:latest
+          crane copy 369495373322.dkr.ecr.eu-central-1.amazonaws.com/neon:$GITHUB_RUN_ID 093970136003.dkr.ecr.us-east-2.amazonaws.com/neon:latest
+          crane copy 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-tools:$GITHUB_RUN_ID 093970136003.dkr.ecr.us-east-2.amazonaws.com/compute-tools:latest
+          crane copy 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node:$GITHUB_RUN_ID 093970136003.dkr.ecr.us-east-2.amazonaws.com/compute-node:latest
 
       - name: Configure Docker Hub login
         run: |
@@ -677,12 +669,12 @@ jobs:
       - id: set-matrix
         run: |
           if [[ "$GITHUB_REF_NAME" == "main" ]]; then
-            STAGING='{"env_name": "staging", "proxy_job": "neon-proxy", "proxy_config": "staging.proxy", "kubeconfig_secret": "STAGING_KUBECONFIG_DATA", "console_api_key_secret": "NEON_STAGING_API_KEY"}'
-            NEON_STRESS='{"env_name": "neon-stress", "proxy_job": "neon-stress-proxy", "proxy_config": "neon-stress.proxy", "kubeconfig_secret": "NEON_STRESS_KUBECONFIG_DATA", "console_api_key_secret": "NEON_CAPTEST_API_KEY"}'
-            echo "include=[$STAGING, $NEON_STRESS]" >> $GITHUB_OUTPUT
+            STAGING='{"env_name": "staging", "proxy_job": "neon-proxy", "proxy_config": "staging.proxy", "kubeconfig_secret": "STAGING_KUBECONFIG_DATA"}'
+            NEON_STRESS='{"env_name": "neon-stress", "proxy_job": "neon-stress-proxy", "proxy_config": "neon-stress.proxy", "kubeconfig_secret": "NEON_STRESS_KUBECONFIG_DATA"}'
+            echo "::set-output name=include::[$STAGING, $NEON_STRESS]"
           elif [[ "$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"}'
-            echo "include=[$PRODUCTION]" >> $GITHUB_OUTPUT
+            PRODUCTION='{"env_name": "production", "proxy_job": "neon-proxy", "proxy_config": "production.proxy", "kubeconfig_secret": "PRODUCTION_KUBECONFIG_DATA"}'
+            echo "::set-output name=include::[$PRODUCTION]"
           else
             echo "GITHUB_REF_NAME (value '$GITHUB_REF_NAME') is not set to either 'main' or 'release'"
             exit 1
@@ -718,7 +710,7 @@ jobs:
       - name: Setup ansible
         run: |
           export PATH="/root/.local/bin:$PATH"
-          pip install --progress-bar off --user ansible boto3 toml
+          pip install --progress-bar off --user ansible boto3
 
       - name: Redeploy
         run: |
@@ -740,48 +732,8 @@ jobs:
           chmod 0600 ssh-key
           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] }}
-          rm -f neon_install.tar.gz .neon_current_version
 
-  deploy-new:
-    runs-on: dev
-    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, calculate-deploy-targets, tag, regress-tests ]
-    if: |
-      (github.ref_name == 'main') &&
-      github.event_name != 'workflow_dispatch'
-    defaults:
-      run:
-        shell: bash
-    env:
-      AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_DEV }}
-      AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_KEY_DEV }}
-    steps:
-      - name: Checkout
-        uses: actions/checkout@v3
-        with:
-          submodules: true
-          fetch-depth: 0
-
-      - name: Redeploy
-        run: |
-          export DOCKER_TAG=${{needs.tag.outputs.build-tag}}
-          cd "$(pwd)/.github/ansible"
-
-          if [[ "$GITHUB_REF_NAME" == "main" ]]; then
-            ./get_binaries.sh
-          elif [[ "$GITHUB_REF_NAME" == "release" ]]; then
-            RELEASE=true ./get_binaries.sh
-          else
-            echo "GITHUB_REF_NAME (value '$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.us-east-2.hosts.yaml -e @ssm_config -e CONSOLE_API_TOKEN=${{secrets.NEON_STAGING_API_KEY}}
+          ansible-playbook deploy.yaml -i ${{ matrix.env_name }}.hosts
           rm -f neon_install.tar.gz .neon_current_version
 
   deploy-proxy:
@@ -825,52 +777,3 @@ jobs:
           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=${DOCKER_TAG} --wait --timeout 15m0s
-
-  deploy-proxy-new:
-    runs-on: dev
-    container: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/ansible:pinned
-    # Compute image isn't strictly required for proxy deploy, but let's still wait for it to run all deploy jobs consistently.
-    needs: [ push-docker-hub, calculate-deploy-targets, tag, regress-tests ]
-    if: |
-      (github.ref_name == 'main') &&
-      github.event_name != 'workflow_dispatch'
-    defaults:
-      run:
-        shell: bash
-    steps:
-      - name: Checkout
-        uses: actions/checkout@v3
-        with:
-          submodules: true
-          fetch-depth: 0
-
-      - name: Configure environment
-        run: |
-          helm repo add neondatabase https://neondatabase.github.io/helm-charts
-          aws --region us-east-2 eks update-kubeconfig --name dev-us-east-2-beta --role-arn arn:aws:iam::369495373322:role/github-runner
-
-      - name: Re-deploy proxy
-        run: |
-          DOCKER_TAG=${{needs.tag.outputs.build-tag}}
-          helm upgrade neon-proxy-scram neondatabase/neon-proxy --namespace neon-proxy --create-namespace --install -f .github/helm-values/dev-us-east-2-beta.neon-proxy-scram.yaml --set image.tag=${DOCKER_TAG} --wait --timeout 15m0s
-
-  promote-compatibility-test-snapshot:
-    runs-on: dev
-    container:
-      image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned
-      options: --init
-    needs: [ deploy, deploy-proxy ]
-    if: github.ref_name == 'release' && github.event_name != 'workflow_dispatch'
-    steps:
-      - name: Promote compatibility snapshot for the release
-        shell: bash -euxo pipefail {0}
-        env:
-          BUCKET: neon-github-public-dev
-          PREFIX: artifacts/latest
-        run: |
-          for build_type in debug release; do
-            OLD_FILENAME=compatibility-snapshot-${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

.github/workflows/codestyle.yml

@@ -36,7 +36,7 @@ jobs:
 
     steps:
       - name: Checkout
-        uses: actions/checkout@v3
+        uses: actions/checkout@v2
         with:
           submodules: true
           fetch-depth: 2
@@ -56,12 +56,12 @@ jobs:
 
       - name: Set pg 14 revision for caching
         id: pg_v14_rev
-        run: echo pg_rev=$(git rev-parse HEAD:vendor/postgres-v14) >> $GITHUB_OUTPUT
+        run: echo ::set-output name=pg_rev::$(git rev-parse HEAD:vendor/postgres-v14)
         shell: bash -euxo pipefail {0}
 
       - name: Set pg 15 revision for caching
         id: pg_v15_rev
-        run: echo pg_rev=$(git rev-parse HEAD:vendor/postgres-v15) >> $GITHUB_OUTPUT
+        run: echo ::set-output name=pg_rev::$(git rev-parse HEAD:vendor/postgres-v15)
         shell: bash -euxo pipefail {0}
 
       - name: Cache postgres v14 build

Cargo.toml

@@ -1,14 +1,3 @@
-# 'named-profiles' feature was stabilized in cargo 1.57. This line makes the
-# build work with older cargo versions.
-#
-# We have this because as of this writing, the latest cargo Debian package
-# that's available is 1.56. (Confusingly, the Debian package version number
-# is 0.57, whereas 'cargo --version' says 1.56.)
-#
-# See https://tracker.debian.org/pkg/cargo for the current status of the
-# package. When that gets updated, we can remove this.
-cargo-features = ["named-profiles"]
-
 [workspace]
 members = [
     "compute_tools",

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 safekeeper --bin proxy --locked --release \
     && cachepot -s
 
 # Build final image
@@ -65,7 +65,6 @@ RUN set -e \
 
 COPY --from=build --chown=neon:neon /home/nonroot/target/release/pageserver          /usr/local/bin
 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/proxy               /usr/local/bin
 

Dockerfile.compute-node-v14

@@ -1,26 +1,24 @@
-#
-# This file is identical to the Dockerfile.compute-node-v15 file
-# except for the version of Postgres that is built.
-#
-
 ARG TAG=pinned
+# apparently, ARGs don't get replaced in RUN commands in kaniko
+# ARG POSTGIS_VERSION=3.3.0
+# ARG PLV8_VERSION=3.1.4
+# ARG PG_VERSION=v14
 
-#########################################################################################
 #
 # Layer "build-deps"
 #
-#########################################################################################
 FROM debian:bullseye-slim AS build-deps
+RUN echo "deb http://ftp.debian.org/debian testing main" >> /etc/apt/sources.list && \
+    echo "APT::Default-Release \"stable\";" > /etc/apt/apt.conf.d/default-release && \
+    apt update
 RUN apt update &&  \
-    apt install -y git autoconf automake libtool build-essential bison flex libreadline-dev \
-    zlib1g-dev libxml2-dev libcurl4-openssl-dev libossp-uuid-dev wget pkg-config
+    apt install -y git autoconf automake libtool build-essential bison flex libreadline-dev zlib1g-dev libxml2-dev \
+    libcurl4-openssl-dev libossp-uuid-dev wget pkg-config libglib2.0-dev
 
-#########################################################################################
 #
 # Layer "pg-build"
 # Build Postgres from the neon postgres repository.
 #
-#########################################################################################
 FROM build-deps AS pg-build
 COPY vendor/postgres-v14 postgres
 RUN cd postgres && \
@@ -31,20 +29,22 @@ RUN cd postgres && \
     make MAKELEVEL=0 -j $(getconf _NPROCESSORS_ONLN) -s -C src/include install && \
     make MAKELEVEL=0 -j $(getconf _NPROCESSORS_ONLN) -s -C src/interfaces/libpq install
 
-#########################################################################################
 #
 # Layer "postgis-build"
 # Build PostGIS from the upstream PostGIS mirror.
 #
-#########################################################################################
+# PostGIS compiles against neon postgres sources without changes. Perhaps we
+# could even use the upstream binaries, compiled against vanilla Postgres, but
+# it would require some investigation to check that it works, and also keeps
+# working in the future. So for now, we compile our own binaries.
 FROM build-deps AS postgis-build
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 RUN apt update && \
     apt install -y gdal-bin libgdal-dev libprotobuf-c-dev protobuf-c-compiler xsltproc
 
-RUN wget https://download.osgeo.org/postgis/source/postgis-3.3.1.tar.gz && \
-    tar xvzf postgis-3.3.1.tar.gz && \
-    cd postgis-3.3.1 && \
+RUN wget https://download.osgeo.org/postgis/source/postgis-3.3.0.tar.gz && \
+    tar xvzf postgis-3.3.0.tar.gz && \
+    cd postgis-3.3.0 && \
     ./autogen.sh && \
     export PATH="/usr/local/pgsql/bin:$PATH" && \
     ./configure && \
@@ -57,55 +57,39 @@ RUN wget https://download.osgeo.org/postgis/source/postgis-3.3.1.tar.gz && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/postgis_tiger_geocoder.control && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/postgis_topology.control
 
-#########################################################################################
 #
 # Layer "plv8-build"
 # Build plv8
 #
-#########################################################################################
 FROM build-deps AS plv8-build
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 RUN apt update && \
-    apt install -y ninja-build python3-dev libc++-dev libc++abi-dev libncurses5 binutils
+    apt install -y ninja-build python3-dev libc++-dev libc++abi-dev libncurses5
 
-# https://github.com/plv8/plv8/issues/475:
-#   v8 uses gold for linking and sets `--thread-count=4` which breaks
-#   gold version <= 1.35 (https://sourceware.org/bugzilla/show_bug.cgi?id=23607)
-# Install newer gold version manually as debian-testing binutils version updates
-# libc version, which in turn breaks other extension built against non-testing libc.
-RUN wget https://ftp.gnu.org/gnu/binutils/binutils-2.38.tar.gz && \
-    tar xvzf binutils-2.38.tar.gz && \
-    cd binutils-2.38 && \
-    cd libiberty && ./configure && make -j $(getconf _NPROCESSORS_ONLN) && \
-    cd ../bfd && ./configure && make bfdver.h && \
-    cd ../gold && ./configure && make -j $(getconf _NPROCESSORS_ONLN) && make install && \
-    cp /usr/local/bin/ld.gold /usr/bin/gold
+# https://github.com/plv8/plv8/issues/475
+# Debian bullseye provides binutils 2.35 when >= 2.38 is necessary
+RUN apt update && \
+    apt install -y --no-install-recommends -t testing binutils
 
-# Sed is used to patch for https://github.com/plv8/plv8/issues/503
 RUN wget https://github.com/plv8/plv8/archive/refs/tags/v3.1.4.tar.gz && \
     tar xvzf v3.1.4.tar.gz && \
     cd plv8-3.1.4 && \
     export PATH="/usr/local/pgsql/bin:$PATH" && \
-    sed -i 's/MemoryContextAlloc(/MemoryContextAllocZero(/' plv8.cc && \
-    make DOCKER=1 -j $(getconf _NPROCESSORS_ONLN) install && \
+    make -j $(getconf _NPROCESSORS_ONLN) && \
+    make -j $(getconf _NPROCESSORS_ONLN) install && \
     rm -rf /plv8-* && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/plv8.control
 
-#########################################################################################
 #
 # Layer "h3-pg-build"
 # Build h3_pg
 #
-#########################################################################################
 FROM build-deps AS h3-pg-build
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
 # packaged cmake is too old
-RUN wget https://github.com/Kitware/CMake/releases/download/v3.24.2/cmake-3.24.2-linux-x86_64.sh \
-      -q -O /tmp/cmake-install.sh \
-      && chmod u+x /tmp/cmake-install.sh \
-      && /tmp/cmake-install.sh --skip-license --prefix=/usr/local/ \
-      && rm /tmp/cmake-install.sh
+RUN apt update && \
+    apt install -y --no-install-recommends -t testing cmake
 
 RUN wget https://github.com/uber/h3/archive/refs/tags/v4.0.1.tar.gz -O h3.tgz && \
     tar xvzf h3.tgz  && \
@@ -124,18 +108,16 @@ RUN wget https://github.com/zachasme/h3-pg/archive/refs/tags/v4.0.1.tar.gz -O h3
     export PATH="/usr/local/pgsql/bin:$PATH" && \
     make -j $(getconf _NPROCESSORS_ONLN) && \
     make -j $(getconf _NPROCESSORS_ONLN) install && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/h3.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/h3_postgis.control
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/h3.control
 
-#########################################################################################
 #
 # Layer "neon-pg-ext-build"
 # compile neon extensions
 #
-#########################################################################################
 FROM build-deps AS neon-pg-ext-build
 COPY --from=postgis-build /usr/local/pgsql/ /usr/local/pgsql/
-COPY --from=plv8-build /usr/local/pgsql/ /usr/local/pgsql/
+# plv8 still sometimes crashes during the creation
+# COPY --from=plv8-build /usr/local/pgsql/ /usr/local/pgsql/
 COPY --from=h3-pg-build /usr/local/pgsql/ /usr/local/pgsql/
 COPY --from=h3-pg-build /h3/usr /
 COPY pgxn/ pgxn/
@@ -145,22 +127,16 @@ RUN make -j $(getconf _NPROCESSORS_ONLN) \
         -C pgxn/neon \
         -s install
 
-#########################################################################################
-#
 # Compile and run the Neon-specific `compute_ctl` binary
-#
-#########################################################################################
 FROM 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:$TAG AS compute-tools
 USER nonroot
 # Copy entire project to get Cargo.* files with proper dependencies for the whole project
 COPY --chown=nonroot . .
 RUN cd compute_tools && cargo build --locked --profile release-line-debug-size-lto
 
-#########################################################################################
 #
 # Clean up postgres folder before inclusion
 #
-#########################################################################################
 FROM neon-pg-ext-build AS postgres-cleanup-layer
 COPY --from=neon-pg-ext-build /usr/local/pgsql /usr/local/pgsql
 
@@ -178,12 +154,10 @@ RUN rm -r /usr/local/pgsql/lib/pgxs/src
 # if they were to be used by other libraries.
 RUN rm /usr/local/pgsql/lib/lib*.a
 
-#########################################################################################
 #
 # Final layer
 # Put it all together into the final image
 #
-#########################################################################################
 FROM debian:bullseye-slim
 # Add user postgres
 RUN mkdir /var/db && useradd -m -d /var/db/postgres postgres && \
@@ -200,6 +174,8 @@ 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
+# GLIBC 2.34 for plv8.
+#     Debian bullseye provides GLIBC 2.31, so we install the library from testing
 #
 # Lastly, link compute_ctl into zenith_ctl while we're at it,
 # so that we don't need to put this in another layer.
@@ -212,6 +188,12 @@ RUN apt update &&  \
         libproj19 \
         libprotobuf-c1 && \
     rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/* && \
+    echo "Installing GLIBC 2.34" && \
+    echo "deb http://ftp.debian.org/debian testing main" >> /etc/apt/sources.list && \
+    echo "APT::Default-Release \"stable\";" > /etc/apt/apt.conf.d/default-release && \
+    apt update && \
+    apt install -y --no-install-recommends -t testing libc6 && \
+    rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/* && \
     ln /usr/local/bin/compute_ctl /usr/local/bin/zenith_ctl
 
 USER postgres

Dockerfile.compute-node-v15

@@ -4,23 +4,26 @@
 #
 
 ARG TAG=pinned
+# apparently, ARGs don't get replaced in RUN commands in kaniko
+# ARG POSTGIS_VERSION=3.3.1
+# ARG PLV8_VERSION=3.1.4
+# ARG PG_VERSION=v15
 
-#########################################################################################
 #
 # Layer "build-deps"
 #
-#########################################################################################
 FROM debian:bullseye-slim AS build-deps
+RUN echo "deb http://ftp.debian.org/debian testing main" >> /etc/apt/sources.list && \
+    echo "APT::Default-Release \"stable\";" > /etc/apt/apt.conf.d/default-release && \
+    apt update
 RUN apt update &&  \
-    apt install -y git autoconf automake libtool build-essential bison flex libreadline-dev \
-    zlib1g-dev libxml2-dev libcurl4-openssl-dev libossp-uuid-dev wget pkg-config
+    apt install -y git autoconf automake libtool build-essential bison flex libreadline-dev zlib1g-dev libxml2-dev \
+    libcurl4-openssl-dev libossp-uuid-dev wget pkg-config libglib2.0-dev
 
-#########################################################################################
 #
 # Layer "pg-build"
 # Build Postgres from the neon postgres repository.
 #
-#########################################################################################
 FROM build-deps AS pg-build
 COPY vendor/postgres-v15 postgres
 RUN cd postgres && \
@@ -31,12 +34,14 @@ RUN cd postgres && \
     make MAKELEVEL=0 -j $(getconf _NPROCESSORS_ONLN) -s -C src/include install && \
     make MAKELEVEL=0 -j $(getconf _NPROCESSORS_ONLN) -s -C src/interfaces/libpq install
 
-#########################################################################################
 #
 # Layer "postgis-build"
 # Build PostGIS from the upstream PostGIS mirror.
 #
-#########################################################################################
+# PostGIS compiles against neon postgres sources without changes. Perhaps we
+# could even use the upstream binaries, compiled against vanilla Postgres, but
+# it would require some investigation to check that it works, and also keeps
+# working in the future. So for now, we compile our own binaries.
 FROM build-deps AS postgis-build
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 RUN apt update && \
@@ -57,55 +62,39 @@ RUN wget https://download.osgeo.org/postgis/source/postgis-3.3.1.tar.gz && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/postgis_tiger_geocoder.control && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/postgis_topology.control
 
-#########################################################################################
 #
 # Layer "plv8-build"
 # Build plv8
 #
-#########################################################################################
 FROM build-deps AS plv8-build
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 RUN apt update && \
-    apt install -y ninja-build python3-dev libc++-dev libc++abi-dev libncurses5 binutils
+    apt install -y ninja-build python3-dev libc++-dev libc++abi-dev libncurses5
 
-# https://github.com/plv8/plv8/issues/475:
-#   v8 uses gold for linking and sets `--thread-count=4` which breaks
-#   gold version <= 1.35 (https://sourceware.org/bugzilla/show_bug.cgi?id=23607)
-# Install newer gold version manually as debian-testing binutils version updates
-# libc version, which in turn breaks other extension built against non-testing libc.
-RUN wget https://ftp.gnu.org/gnu/binutils/binutils-2.38.tar.gz && \
-    tar xvzf binutils-2.38.tar.gz && \
-    cd binutils-2.38 && \
-    cd libiberty && ./configure && make -j $(getconf _NPROCESSORS_ONLN) && \
-    cd ../bfd && ./configure && make bfdver.h && \
-    cd ../gold && ./configure && make -j $(getconf _NPROCESSORS_ONLN) && make install && \
-    cp /usr/local/bin/ld.gold /usr/bin/gold
+# https://github.com/plv8/plv8/issues/475
+# Debian bullseye provides binutils 2.35 when >= 2.38 is necessary
+RUN apt update && \
+    apt install -y --no-install-recommends -t testing binutils
 
-# Sed is used to patch for https://github.com/plv8/plv8/issues/503
 RUN wget https://github.com/plv8/plv8/archive/refs/tags/v3.1.4.tar.gz && \
     tar xvzf v3.1.4.tar.gz && \
     cd plv8-3.1.4 && \
     export PATH="/usr/local/pgsql/bin:$PATH" && \
-    sed -i 's/MemoryContextAlloc(/MemoryContextAllocZero(/' plv8.cc && \
-    make DOCKER=1 -j $(getconf _NPROCESSORS_ONLN) install && \
+    make -j $(getconf _NPROCESSORS_ONLN) && \
+    make -j $(getconf _NPROCESSORS_ONLN) install && \
     rm -rf /plv8-* && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/plv8.control
 
-#########################################################################################
 #
 # Layer "h3-pg-build"
 # Build h3_pg
 #
-#########################################################################################
 FROM build-deps AS h3-pg-build
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
 # packaged cmake is too old
-RUN wget https://github.com/Kitware/CMake/releases/download/v3.24.2/cmake-3.24.2-linux-x86_64.sh \
-      -q -O /tmp/cmake-install.sh \
-      && chmod u+x /tmp/cmake-install.sh \
-      && /tmp/cmake-install.sh --skip-license --prefix=/usr/local/ \
-      && rm /tmp/cmake-install.sh
+RUN apt update && \
+    apt install -y --no-install-recommends -t testing cmake
 
 RUN wget https://github.com/uber/h3/archive/refs/tags/v4.0.1.tar.gz -O h3.tgz && \
     tar xvzf h3.tgz  && \
@@ -124,18 +113,16 @@ RUN wget https://github.com/zachasme/h3-pg/archive/refs/tags/v4.0.1.tar.gz -O h3
     export PATH="/usr/local/pgsql/bin:$PATH" && \
     make -j $(getconf _NPROCESSORS_ONLN) && \
     make -j $(getconf _NPROCESSORS_ONLN) install && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/h3.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/h3_postgis.control
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/h3.control
 
-#########################################################################################
 #
 # Layer "neon-pg-ext-build"
 # compile neon extensions
 #
-#########################################################################################
 FROM build-deps AS neon-pg-ext-build
 COPY --from=postgis-build /usr/local/pgsql/ /usr/local/pgsql/
-COPY --from=plv8-build /usr/local/pgsql/ /usr/local/pgsql/
+# plv8 still sometimes crashes during the creation
+# COPY --from=plv8-build /usr/local/pgsql/ /usr/local/pgsql/
 COPY --from=h3-pg-build /usr/local/pgsql/ /usr/local/pgsql/
 COPY --from=h3-pg-build /h3/usr /
 COPY pgxn/ pgxn/
@@ -145,22 +132,16 @@ RUN make -j $(getconf _NPROCESSORS_ONLN) \
         -C pgxn/neon \
         -s install
 
-#########################################################################################
-#
 # Compile and run the Neon-specific `compute_ctl` binary
-#
-#########################################################################################
 FROM 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:$TAG AS compute-tools
 USER nonroot
 # Copy entire project to get Cargo.* files with proper dependencies for the whole project
 COPY --chown=nonroot . .
 RUN cd compute_tools && cargo build --locked --profile release-line-debug-size-lto
 
-#########################################################################################
 #
 # Clean up postgres folder before inclusion
 #
-#########################################################################################
 FROM neon-pg-ext-build AS postgres-cleanup-layer
 COPY --from=neon-pg-ext-build /usr/local/pgsql /usr/local/pgsql
 
@@ -178,12 +159,10 @@ RUN rm -r /usr/local/pgsql/lib/pgxs/src
 # if they were to be used by other libraries.
 RUN rm /usr/local/pgsql/lib/lib*.a
 
-#########################################################################################
 #
 # Final layer
 # Put it all together into the final image
 #
-#########################################################################################
 FROM debian:bullseye-slim
 # Add user postgres
 RUN mkdir /var/db && useradd -m -d /var/db/postgres postgres && \
@@ -200,6 +179,8 @@ 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
+# GLIBC 2.34 for plv8.
+#     Debian bullseye provides GLIBC 2.31, so we install the library from testing
 #
 # Lastly, link compute_ctl into zenith_ctl while we're at it,
 # so that we don't need to put this in another layer.
@@ -212,6 +193,12 @@ RUN apt update &&  \
         libproj19 \
         libprotobuf-c1 && \
     rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/* && \
+    echo "Installing GLIBC 2.34" && \
+    echo "deb http://ftp.debian.org/debian testing main" >> /etc/apt/sources.list && \
+    echo "APT::Default-Release \"stable\";" > /etc/apt/apt.conf.d/default-release && \
+    apt update && \
+    apt install -y --no-install-recommends -t testing libc6 && \
+    rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/* && \
     ln /usr/local/bin/compute_ctl /usr/local/bin/zenith_ctl
 
 USER postgres

compute_tools/Cargo.toml

@@ -6,7 +6,7 @@ edition = "2021"
 [dependencies]
 anyhow = "1.0"
 chrono = "0.4"
-clap = "4.0"
+clap = "3.0"
 env_logger = "0.9"
 futures = "0.3.13"
 hyper = { version = "0.14", features = ["full"] }

compute_tools/src/bin/compute_ctl.rs

@@ -51,19 +51,53 @@ fn main() -> Result<()> {
     // TODO: re-use `utils::logging` later
     init_logger(DEFAULT_LOG_LEVEL)?;
 
-    let matches = cli().get_matches();
+    // Env variable is set by `cargo`
+    let version: Option<&str> = option_env!("CARGO_PKG_VERSION");
+    let matches = clap::App::new("compute_ctl")
+        .version(version.unwrap_or("unknown"))
+        .arg(
+            Arg::new("connstr")
+                .short('C')
+                .long("connstr")
+                .value_name("DATABASE_URL")
+                .required(true),
+        )
+        .arg(
+            Arg::new("pgdata")
+                .short('D')
+                .long("pgdata")
+                .value_name("DATADIR")
+                .required(true),
+        )
+        .arg(
+            Arg::new("pgbin")
+                .short('b')
+                .long("pgbin")
+                .value_name("POSTGRES_PATH"),
+        )
+        .arg(
+            Arg::new("spec")
+                .short('s')
+                .long("spec")
+                .value_name("SPEC_JSON"),
+        )
+        .arg(
+            Arg::new("spec-path")
+                .short('S')
+                .long("spec-path")
+                .value_name("SPEC_PATH"),
+        )
+        .get_matches();
 
-    let pgdata = matches
-        .get_one::<String>("pgdata")
-        .expect("PGDATA path is required");
+    let pgdata = matches.value_of("pgdata").expect("PGDATA path is required");
     let connstr = matches
-        .get_one::<String>("connstr")
+        .value_of("connstr")
         .expect("Postgres connection string is required");
-    let spec = matches.get_one::<String>("spec");
-    let spec_path = matches.get_one::<String>("spec-path");
+    let spec = matches.value_of("spec");
+    let spec_path = matches.value_of("spec-path");
 
     // Try to use just 'postgres' if no path is provided
-    let pgbin = matches.get_one::<String>("pgbin").unwrap();
+    let pgbin = matches.value_of("pgbin").unwrap_or("postgres");
 
     let spec: ComputeSpec = match spec {
         // First, try to get cluster spec from the cli argument
@@ -139,48 +173,3 @@ fn main() -> Result<()> {
         }
     }
 }
-
-fn cli() -> clap::Command {
-    // Env variable is set by `cargo`
-    let version = option_env!("CARGO_PKG_VERSION").unwrap_or("unknown");
-    clap::Command::new("compute_ctl")
-        .version(version)
-        .arg(
-            Arg::new("connstr")
-                .short('C')
-                .long("connstr")
-                .value_name("DATABASE_URL")
-                .required(true),
-        )
-        .arg(
-            Arg::new("pgdata")
-                .short('D')
-                .long("pgdata")
-                .value_name("DATADIR")
-                .required(true),
-        )
-        .arg(
-            Arg::new("pgbin")
-                .short('b')
-                .long("pgbin")
-                .default_value("postgres")
-                .value_name("POSTGRES_PATH"),
-        )
-        .arg(
-            Arg::new("spec")
-                .short('s')
-                .long("spec")
-                .value_name("SPEC_JSON"),
-        )
-        .arg(
-            Arg::new("spec-path")
-                .short('S')
-                .long("spec-path")
-                .value_name("SPEC_PATH"),
-        )
-}
-
-#[test]
-fn verify_cli() {
-    cli().debug_assert()
-}

compute_tools/src/pg_helpers.rs

@@ -8,10 +8,11 @@ use std::process::Child;
 use std::time::{Duration, Instant};
 
 use anyhow::{bail, Result};
-use notify::{RecursiveMode, Watcher};
 use postgres::{Client, Transaction};
 use serde::Deserialize;
 
+use notify::{RecursiveMode, Watcher};
+
 const POSTGRES_WAIT_TIMEOUT: Duration = Duration::from_millis(60 * 1000); // milliseconds
 
 /// Rust representation of Postgres role info with only those fields
@@ -168,7 +169,7 @@ impl Database {
     /// it may require a proper quoting too.
     pub fn to_pg_options(&self) -> String {
         let mut params: String = self.options.as_pg_options();
-        write!(params, " OWNER {}", &self.owner.pg_quote())
+        write!(params, " OWNER {}", &self.owner.quote())
             .expect("String is documented to not to error during write operations");
 
         params
@@ -179,17 +180,18 @@ impl Database {
 /// intended to be used for DB / role names.
 pub type PgIdent = String;
 
-/// Generic trait used to provide quoting / encoding for strings used in the
-/// Postgres SQL queries and DATABASE_URL.
-pub trait Escaping {
-    fn pg_quote(&self) -> String;
+/// Generic trait used to provide quoting for strings used in the
+/// Postgres SQL queries. Currently used only to implement quoting
+/// of identifiers, but could be used for literals in the future.
+pub trait PgQuote {
+    fn quote(&self) -> String;
 }
 
-impl Escaping for PgIdent {
+impl PgQuote for PgIdent {
     /// This is intended to mimic Postgres quote_ident(), but for simplicity it
-    /// always quotes provided string with `""` and escapes every `"`.
-    /// **Not idempotent**, i.e. if string is already escaped it will be escaped again.
-    fn pg_quote(&self) -> String {
+    /// always quotes provided string with `""` and escapes every `"`. Not idempotent,
+    /// i.e. if string is already escaped it will be escaped again.
+    fn quote(&self) -> String {
         let result = format!("\"{}\"", self.replace('"', "\"\""));
         result
     }

compute_tools/src/spec.rs

@@ -1,9 +1,7 @@
 use std::path::Path;
-use std::str::FromStr;
 
 use anyhow::Result;
 use log::{info, log_enabled, warn, Level};
-use postgres::config::Config;
 use postgres::{Client, NoTls};
 use serde::Deserialize;
 
@@ -117,8 +115,8 @@ pub fn handle_roles(spec: &ComputeSpec, client: &mut Client) -> Result<()> {
                     if existing_roles.iter().any(|r| r.name == op.name) {
                         let query: String = format!(
                             "ALTER ROLE {} RENAME TO {}",
-                            op.name.pg_quote(),
-                            new_name.pg_quote()
+                            op.name.quote(),
+                            new_name.quote()
                         );
 
                         warn!("renaming role '{}' to '{}'", op.name, new_name);
@@ -164,7 +162,7 @@ pub fn handle_roles(spec: &ComputeSpec, client: &mut Client) -> Result<()> {
             }
 
             if update_role {
-                let mut query: String = format!("ALTER ROLE {} ", name.pg_quote());
+                let mut query: String = format!("ALTER ROLE {} ", name.quote());
                 info_print!(" -> update");
 
                 query.push_str(&role.to_pg_options());
@@ -172,7 +170,7 @@ pub fn handle_roles(spec: &ComputeSpec, client: &mut Client) -> Result<()> {
             }
         } else {
             info!("role name: '{}'", &name);
-            let mut query: String = format!("CREATE ROLE {} ", name.pg_quote());
+            let mut query: String = format!("CREATE ROLE {} ", name.quote());
             info!("role create query: '{}'", &query);
             info_print!(" -> create");
 
@@ -181,7 +179,7 @@ pub fn handle_roles(spec: &ComputeSpec, client: &mut Client) -> Result<()> {
 
             let grant_query = format!(
                 "GRANT pg_read_all_data, pg_write_all_data TO {}",
-                name.pg_quote()
+                name.quote()
             );
             xact.execute(grant_query.as_str(), &[])?;
             info!("role grant query: '{}'", &grant_query);
@@ -217,7 +215,7 @@ pub fn handle_role_deletions(node: &ComputeNode, client: &mut Client) -> Result<
             // We do not check either role exists or not,
             // Postgres will take care of it for us
             if op.action == "delete_role" {
-                let query: String = format!("DROP ROLE IF EXISTS {}", &op.name.pg_quote());
+                let query: String = format!("DROP ROLE IF EXISTS {}", &op.name.quote());
 
                 warn!("deleting role '{}'", &op.name);
                 xact.execute(query.as_str(), &[])?;
@@ -232,16 +230,17 @@ pub fn handle_role_deletions(node: &ComputeNode, client: &mut Client) -> Result<
 fn reassign_owned_objects(node: &ComputeNode, role_name: &PgIdent) -> Result<()> {
     for db in &node.spec.cluster.databases {
         if db.owner != *role_name {
-            let mut conf = Config::from_str(node.connstr.as_str())?;
-            conf.dbname(&db.name);
+            let mut connstr = node.connstr.clone();
+            // database name is always the last and the only component of the path
+            connstr.set_path(&db.name);
 
-            let mut client = conf.connect(NoTls)?;
+            let mut client = Client::connect(connstr.as_str(), NoTls)?;
 
             // This will reassign all dependent objects to the db owner
             let reassign_query = format!(
                 "REASSIGN OWNED BY {} TO {}",
-                role_name.pg_quote(),
-                db.owner.pg_quote()
+                role_name.quote(),
+                db.owner.quote()
             );
             info!(
                 "reassigning objects owned by '{}' in db '{}' to '{}'",
@@ -250,7 +249,7 @@ fn reassign_owned_objects(node: &ComputeNode, role_name: &PgIdent) -> Result<()>
             client.simple_query(&reassign_query)?;
 
             // This now will only drop privileges of the role
-            let drop_query = format!("DROP OWNED BY {}", role_name.pg_quote());
+            let drop_query = format!("DROP OWNED BY {}", role_name.quote());
             client.simple_query(&drop_query)?;
         }
     }
@@ -280,7 +279,7 @@ pub fn handle_databases(spec: &ComputeSpec, client: &mut Client) -> Result<()> {
                 // We do not check either DB exists or not,
                 // Postgres will take care of it for us
                 "delete_db" => {
-                    let query: String = format!("DROP DATABASE IF EXISTS {}", &op.name.pg_quote());
+                    let query: String = format!("DROP DATABASE IF EXISTS {}", &op.name.quote());
 
                     warn!("deleting database '{}'", &op.name);
                     client.execute(query.as_str(), &[])?;
@@ -292,8 +291,8 @@ pub fn handle_databases(spec: &ComputeSpec, client: &mut Client) -> Result<()> {
                     if existing_dbs.iter().any(|r| r.name == op.name) {
                         let query: String = format!(
                             "ALTER DATABASE {} RENAME TO {}",
-                            op.name.pg_quote(),
-                            new_name.pg_quote()
+                            op.name.quote(),
+                            new_name.quote()
                         );
 
                         warn!("renaming database '{}' to '{}'", op.name, new_name);
@@ -321,7 +320,7 @@ pub fn handle_databases(spec: &ComputeSpec, client: &mut Client) -> Result<()> {
             // XXX: db owner name is returned as quoted string from Postgres,
             // when quoting is needed.
             let new_owner = if r.owner.starts_with('"') {
-                db.owner.pg_quote()
+                db.owner.quote()
             } else {
                 db.owner.clone()
             };
@@ -329,15 +328,15 @@ pub fn handle_databases(spec: &ComputeSpec, client: &mut Client) -> Result<()> {
             if new_owner != r.owner {
                 let query: String = format!(
                     "ALTER DATABASE {} OWNER TO {}",
-                    name.pg_quote(),
-                    db.owner.pg_quote()
+                    name.quote(),
+                    db.owner.quote()
                 );
                 info_print!(" -> update");
 
                 client.execute(query.as_str(), &[])?;
             }
         } else {
-            let mut query: String = format!("CREATE DATABASE {} ", name.pg_quote());
+            let mut query: String = format!("CREATE DATABASE {} ", name.quote());
             info_print!(" -> create");
 
             query.push_str(&db.to_pg_options());
@@ -367,7 +366,7 @@ pub fn handle_grants(node: &ComputeNode, client: &mut Client) -> Result<()> {
         .cluster
         .roles
         .iter()
-        .map(|r| r.name.pg_quote())
+        .map(|r| r.name.quote())
         .collect::<Vec<_>>();
 
     for db in &spec.cluster.databases {
@@ -375,7 +374,7 @@ pub fn handle_grants(node: &ComputeNode, client: &mut Client) -> Result<()> {
 
         let query: String = format!(
             "GRANT CREATE ON DATABASE {} TO {}",
-            dbname.pg_quote(),
+            dbname.quote(),
             roles.join(", ")
         );
         info!("grant query {}", &query);
@@ -386,11 +385,12 @@ pub fn handle_grants(node: &ComputeNode, client: &mut Client) -> Result<()> {
     // Do some per-database access adjustments. We'd better do this at db creation time,
     // but CREATE DATABASE isn't transactional. So we cannot create db + do some grants
     // atomically.
+    let mut db_connstr = node.connstr.clone();
     for db in &node.spec.cluster.databases {
-        let mut conf = Config::from_str(node.connstr.as_str())?;
-        conf.dbname(&db.name);
+        // database name is always the last and the only component of the path
+        db_connstr.set_path(&db.name);
 
-        let mut db_client = conf.connect(NoTls)?;
+        let mut db_client = Client::connect(db_connstr.as_str(), NoTls)?;
 
         // This will only change ownership on the schema itself, not the objects
         // inside it. Without it owner of the `public` schema will be `cloud_admin`
@@ -419,36 +419,9 @@ pub fn handle_grants(node: &ComputeNode, client: &mut Client) -> Result<()> {
                     END IF;\n\
                 END\n\
             $$;",
-            db.owner.pg_quote()
+            db.owner.quote()
         );
         db_client.simple_query(&alter_query)?;
-
-        // Explicitly grant CREATE ON SCHEMA PUBLIC to the web_access user.
-        // This is needed because since postgres 15 this privilege is removed by default.
-        let grant_query = "DO $$\n\
-                BEGIN\n\
-                    IF EXISTS(\n\
-                        SELECT nspname\n\
-                        FROM pg_catalog.pg_namespace\n\
-                        WHERE nspname = 'public'\n\
-                    ) AND\n\
-                    current_setting('server_version_num')::int/10000 >= 15\n\
-                    THEN\n\
-                        IF EXISTS(\n\
-                            SELECT rolname\n\
-                            FROM pg_catalog.pg_roles\n\
-                            WHERE rolname = 'web_access'\n\
-                        )\n\
-                        THEN\n\
-                            GRANT CREATE ON SCHEMA public TO web_access;\n\
-                        END IF;\n\
-                    END IF;\n\
-                END\n\
-            $$;"
-        .to_string();
-
-        info!("grant query for db {} : {}", &db.name, &grant_query);
-        db_client.simple_query(&grant_query)?;
     }
 
     Ok(())

compute_tools/tests/pg_helpers_tests.rs

@@ -33,9 +33,9 @@ mod pg_helpers_tests {
     }
 
     #[test]
-    fn ident_pg_quote() {
+    fn quote_ident() {
         let ident: PgIdent = PgIdent::from("\"name\";\\n select 1;");
 
-        assert_eq!(ident.pg_quote(), "\"\"\"name\"\";\\n select 1;\"");
+        assert_eq!(ident.quote(), "\"\"\"name\"\";\\n select 1;\"");
     }
 }

control_plane/Cargo.toml

@@ -4,19 +4,19 @@ version = "0.1.0"
 edition = "2021"
 
 [dependencies]
-clap = "4.0"
-comfy-table = "6.1"
+clap = "3.0"
+comfy-table = "5.0.1"
 git-version = "0.3.5"
 tar = "0.4.38"
-postgres = { git = "https://github.com/neondatabase/rust-postgres.git", rev = "d052ee8b86fff9897c77b0fe89ea9daba0e1fa38" }
+postgres = { git = "https://github.com/neondatabase/rust-postgres.git", rev="d052ee8b86fff9897c77b0fe89ea9daba0e1fa38" }
 serde = { version = "1.0", features = ["derive"] }
-serde_with = "2.0"
+serde_with = "1.12.0"
 toml = "0.5"
 once_cell = "1.13.0"
 regex = "1"
 anyhow = "1.0"
 thiserror = "1"
-nix = "0.25"
+nix = "0.23"
 reqwest = { version = "0.11", default-features = false, features = ["blocking", "json", "rustls-tls"] }
 
 # Note: Do not directly depend on pageserver or safekeeper; use pageserver_api or safekeeper_api

control_plane/src/bin/neon_local.rs

@@ -6,7 +6,7 @@
 //! rely on `neon_local` to set up the environment for each test.
 //!
 use anyhow::{anyhow, bail, Context, Result};
-use clap::{value_parser, Arg, ArgAction, ArgMatches, Command};
+use clap::{App, AppSettings, Arg, ArgMatches};
 use control_plane::compute::ComputeControlPlane;
 use control_plane::local_env::{EtcdBroker, LocalEnv};
 use control_plane::safekeeper::SafekeeperNode;
@@ -85,7 +85,212 @@ struct TimelineTreeEl {
 //   * Providing CLI api to the pageserver
 //   * TODO: export/import to/from usual postgres
 fn main() -> Result<()> {
-    let matches = cli().get_matches();
+    let branch_name_arg = Arg::new("branch-name")
+        .long("branch-name")
+        .takes_value(true)
+        .help("Name of the branch to be created or used as an alias for other services")
+        .required(false);
+
+    let pg_node_arg = Arg::new("node").help("Postgres node name").required(false);
+
+    let safekeeper_id_arg = Arg::new("id").help("safekeeper id").required(false);
+
+    let tenant_id_arg = Arg::new("tenant-id")
+        .long("tenant-id")
+        .help("Tenant id. Represented as a hexadecimal string 32 symbols length")
+        .takes_value(true)
+        .required(false);
+
+    let timeline_id_arg = Arg::new("timeline-id")
+        .long("timeline-id")
+        .help("Timeline id. Represented as a hexadecimal string 32 symbols length")
+        .takes_value(true)
+        .required(false);
+
+    let pg_version_arg = Arg::new("pg-version")
+        .long("pg-version")
+        .help("Postgres version to use for the initial tenant")
+        .required(false)
+        .takes_value(true)
+        .default_value(DEFAULT_PG_VERSION);
+
+    let port_arg = Arg::new("port")
+        .long("port")
+        .required(false)
+        .value_name("port");
+
+    let stop_mode_arg = Arg::new("stop-mode")
+        .short('m')
+        .takes_value(true)
+        .possible_values(&["fast", "immediate"])
+        .help("If 'immediate', don't flush repository data at shutdown")
+        .required(false)
+        .value_name("stop-mode");
+
+    let pageserver_config_args = Arg::new("pageserver-config-override")
+        .long("pageserver-config-override")
+        .takes_value(true)
+        .number_of_values(1)
+        .multiple_occurrences(true)
+        .help("Additional pageserver's configuration options or overrides, refer to pageserver's 'config-override' CLI parameter docs for more")
+        .required(false);
+
+    let lsn_arg = Arg::new("lsn")
+        .long("lsn")
+        .help("Specify Lsn on the timeline to start from. By default, end of the timeline would be used.")
+        .takes_value(true)
+        .required(false);
+
+    let matches = App::new("Neon CLI")
+        .setting(AppSettings::ArgRequiredElseHelp)
+        .version(GIT_VERSION)
+        .subcommand(
+            App::new("init")
+                .about("Initialize a new Neon repository")
+                .arg(pageserver_config_args.clone())
+                .arg(timeline_id_arg.clone().help("Use a specific timeline id when creating a tenant and its initial timeline"))
+                .arg(
+                    Arg::new("config")
+                        .long("config")
+                        .required(false)
+                        .value_name("config"),
+                )
+                .arg(pg_version_arg.clone())
+        )
+        .subcommand(
+            App::new("timeline")
+            .about("Manage timelines")
+            .subcommand(App::new("list")
+                .about("List all timelines, available to this pageserver")
+                .arg(tenant_id_arg.clone()))
+            .subcommand(App::new("branch")
+                .about("Create a new timeline, using another timeline as a base, copying its data")
+                .arg(tenant_id_arg.clone())
+                .arg(branch_name_arg.clone())
+                .arg(Arg::new("ancestor-branch-name").long("ancestor-branch-name").takes_value(true)
+                    .help("Use last Lsn of another timeline (and its data) as base when creating the new timeline. The timeline gets resolved by its branch name.").required(false))
+                .arg(Arg::new("ancestor-start-lsn").long("ancestor-start-lsn").takes_value(true)
+                    .help("When using another timeline as base, use a specific Lsn in it instead of the latest one").required(false)))
+            .subcommand(App::new("create")
+                .about("Create a new blank timeline")
+                .arg(tenant_id_arg.clone())
+                .arg(branch_name_arg.clone())
+                .arg(pg_version_arg.clone())
+            )
+            .subcommand(App::new("import")
+                .about("Import timeline from basebackup directory")
+                .arg(tenant_id_arg.clone())
+                .arg(timeline_id_arg.clone())
+                .arg(Arg::new("node-name").long("node-name").takes_value(true)
+                    .help("Name to assign to the imported timeline"))
+                .arg(Arg::new("base-tarfile").long("base-tarfile").takes_value(true)
+                    .help("Basebackup tarfile to import"))
+                .arg(Arg::new("base-lsn").long("base-lsn").takes_value(true)
+                    .help("Lsn the basebackup starts at"))
+                .arg(Arg::new("wal-tarfile").long("wal-tarfile").takes_value(true)
+                    .help("Wal to add after base"))
+                .arg(Arg::new("end-lsn").long("end-lsn").takes_value(true)
+                    .help("Lsn the basebackup ends at"))
+                .arg(pg_version_arg.clone())
+            )
+        ).subcommand(
+            App::new("tenant")
+            .setting(AppSettings::ArgRequiredElseHelp)
+            .about("Manage tenants")
+            .subcommand(App::new("list"))
+            .subcommand(App::new("create")
+                .arg(tenant_id_arg.clone())
+                .arg(timeline_id_arg.clone().help("Use a specific timeline id when creating a tenant and its initial timeline"))
+                .arg(Arg::new("config").short('c').takes_value(true).multiple_occurrences(true).required(false))
+                .arg(pg_version_arg.clone())
+                )
+            .subcommand(App::new("config")
+                .arg(tenant_id_arg.clone())
+                .arg(Arg::new("config").short('c').takes_value(true).multiple_occurrences(true).required(false))
+                )
+        )
+        .subcommand(
+            App::new("pageserver")
+                .setting(AppSettings::ArgRequiredElseHelp)
+                .about("Manage pageserver")
+                .subcommand(App::new("status"))
+                .subcommand(App::new("start").about("Start local pageserver").arg(pageserver_config_args.clone()))
+                .subcommand(App::new("stop").about("Stop local pageserver")
+                            .arg(stop_mode_arg.clone()))
+                .subcommand(App::new("restart").about("Restart local pageserver").arg(pageserver_config_args.clone()))
+        )
+        .subcommand(
+            App::new("safekeeper")
+                .setting(AppSettings::ArgRequiredElseHelp)
+                .about("Manage safekeepers")
+                .subcommand(App::new("start")
+                            .about("Start local safekeeper")
+                            .arg(safekeeper_id_arg.clone())
+                )
+                .subcommand(App::new("stop")
+                            .about("Stop local safekeeper")
+                            .arg(safekeeper_id_arg.clone())
+                            .arg(stop_mode_arg.clone())
+                )
+                .subcommand(App::new("restart")
+                            .about("Restart local safekeeper")
+                            .arg(safekeeper_id_arg.clone())
+                            .arg(stop_mode_arg.clone())
+                )
+        )
+        .subcommand(
+            App::new("pg")
+                .setting(AppSettings::ArgRequiredElseHelp)
+                .about("Manage postgres instances")
+                .subcommand(App::new("list").arg(tenant_id_arg.clone()))
+                .subcommand(App::new("create")
+                    .about("Create a postgres compute node")
+                    .arg(pg_node_arg.clone())
+                    .arg(branch_name_arg.clone())
+                    .arg(tenant_id_arg.clone())
+                    .arg(lsn_arg.clone())
+                    .arg(port_arg.clone())
+                    .arg(
+                        Arg::new("config-only")
+                            .help("Don't do basebackup, create compute node with only config files")
+                            .long("config-only")
+                            .required(false))
+                    .arg(pg_version_arg.clone())
+                )
+                .subcommand(App::new("start")
+                    .about("Start a postgres compute node.\n This command actually creates new node from scratch, but preserves existing config files")
+                    .arg(pg_node_arg.clone())
+                    .arg(tenant_id_arg.clone())
+                    .arg(branch_name_arg.clone())
+                    .arg(timeline_id_arg.clone())
+                    .arg(lsn_arg.clone())
+                    .arg(port_arg.clone())
+                    .arg(pg_version_arg.clone())
+                )
+                .subcommand(
+                    App::new("stop")
+                    .arg(pg_node_arg.clone())
+                    .arg(tenant_id_arg.clone())
+                    .arg(
+                        Arg::new("destroy")
+                            .help("Also delete data directory (now optional, should be default in future)")
+                            .long("destroy")
+                            .required(false)
+                    )
+                    )
+
+        )
+        .subcommand(
+            App::new("start")
+                .about("Start page server and safekeepers")
+                .arg(pageserver_config_args)
+        )
+        .subcommand(
+            App::new("stop")
+                .about("Stop page server and safekeepers")
+                .arg(stop_mode_arg.clone())
+        )
+        .get_matches();
 
     let (sub_name, sub_args) = match matches.subcommand() {
         Some(subcommand_data) => subcommand_data,
@@ -153,7 +358,9 @@ fn print_timelines_tree(
 
     // Memorize all direct children of each timeline.
     for timeline in timelines.iter() {
-        if let Some(ancestor_timeline_id) = timeline.ancestor_timeline_id {
+        if let Some(ancestor_timeline_id) =
+            timeline.local.as_ref().and_then(|l| l.ancestor_timeline_id)
+        {
             timelines_hash
                 .get_mut(&ancestor_timeline_id)
                 .context("missing timeline info in the HashMap")?
@@ -164,7 +371,13 @@ fn print_timelines_tree(
 
     for timeline in timelines_hash.values() {
         // Start with root local timelines (no ancestors) first.
-        if timeline.info.ancestor_timeline_id.is_none() {
+        if timeline
+            .info
+            .local
+            .as_ref()
+            .and_then(|l| l.ancestor_timeline_id)
+            .is_none()
+        {
             print_timeline(0, &Vec::from([true]), timeline, &timelines_hash)?;
         }
     }
@@ -181,8 +394,17 @@ fn print_timeline(
     timeline: &TimelineTreeEl,
     timelines: &HashMap<TimelineId, TimelineTreeEl>,
 ) -> Result<()> {
+    let local_remote = match (timeline.info.local.as_ref(), timeline.info.remote.as_ref()) {
+        (None, None) => unreachable!("in this case no info for a timeline is found"),
+        (None, Some(_)) => "(R)",
+        (Some(_), None) => "(L)",
+        (Some(_), Some(_)) => "(L+R)",
+    };
+    // Draw main padding
+    print!("{} ", local_remote);
+
     if nesting_level > 0 {
-        let ancestor_lsn = match timeline.info.ancestor_lsn {
+        let ancestor_lsn = match timeline.info.local.as_ref().and_then(|i| i.ancestor_lsn) {
             Some(lsn) => lsn.to_string(),
             None => "Unknown Lsn".to_string(),
         };
@@ -270,16 +492,16 @@ fn get_tenant_id(sub_match: &ArgMatches, env: &local_env::LocalEnv) -> anyhow::R
 
 fn parse_tenant_id(sub_match: &ArgMatches) -> anyhow::Result<Option<TenantId>> {
     sub_match
-        .get_one::<String>("tenant-id")
-        .map(|tenant_id| TenantId::from_str(tenant_id))
+        .value_of("tenant-id")
+        .map(TenantId::from_str)
         .transpose()
         .context("Failed to parse tenant id from the argument string")
 }
 
 fn parse_timeline_id(sub_match: &ArgMatches) -> anyhow::Result<Option<TimelineId>> {
     sub_match
-        .get_one::<String>("timeline-id")
-        .map(|timeline_id| TimelineId::from_str(timeline_id))
+        .value_of("timeline-id")
+        .map(TimelineId::from_str)
         .transpose()
         .context("Failed to parse timeline id from the argument string")
 }
@@ -288,22 +510,19 @@ fn handle_init(init_match: &ArgMatches) -> anyhow::Result<LocalEnv> {
     let initial_timeline_id_arg = parse_timeline_id(init_match)?;
 
     // Create config file
-    let toml_file: String = if let Some(config_path) = init_match.get_one::<PathBuf>("config") {
+    let toml_file: String = if let Some(config_path) = init_match.value_of("config") {
         // load and parse the file
-        std::fs::read_to_string(config_path).with_context(|| {
-            format!(
-                "Could not read configuration file '{}'",
-                config_path.display()
-            )
-        })?
+        std::fs::read_to_string(std::path::Path::new(config_path))
+            .with_context(|| format!("Could not read configuration file '{config_path}'"))?
     } else {
         // Built-in default config
         default_conf(&EtcdBroker::locate_etcd()?)
     };
 
     let pg_version = init_match
-        .get_one::<u32>("pg-version")
-        .copied()
+        .value_of("pg-version")
+        .unwrap()
+        .parse::<u32>()
         .context("Failed to parse postgres version from the argument string")?;
 
     let mut env =
@@ -339,10 +558,9 @@ fn handle_init(init_match: &ArgMatches) -> anyhow::Result<LocalEnv> {
 
 fn pageserver_config_overrides(init_match: &ArgMatches) -> Vec<&str> {
     init_match
-        .get_many::<String>("pageserver-config-override")
+        .values_of("pageserver-config-override")
         .into_iter()
         .flatten()
-        .map(|s| s.as_str())
         .collect()
 }
 
@@ -357,7 +575,7 @@ fn handle_tenant(tenant_match: &ArgMatches, env: &mut local_env::LocalEnv) -> an
         Some(("create", create_match)) => {
             let initial_tenant_id = parse_tenant_id(create_match)?;
             let tenant_conf: HashMap<_, _> = create_match
-                .get_many::<String>("config")
+                .values_of("config")
                 .map(|vals| vals.flat_map(|c| c.split_once(':')).collect())
                 .unwrap_or_default();
             let new_tenant_id = pageserver.tenant_create(initial_tenant_id, tenant_conf)?;
@@ -366,8 +584,9 @@ fn handle_tenant(tenant_match: &ArgMatches, env: &mut local_env::LocalEnv) -> an
             // Create an initial timeline for the new tenant
             let new_timeline_id = parse_timeline_id(create_match)?;
             let pg_version = create_match
-                .get_one::<u32>("pg-version")
-                .copied()
+                .value_of("pg-version")
+                .unwrap()
+                .parse::<u32>()
                 .context("Failed to parse postgres version from the argument string")?;
 
             let timeline_info = pageserver.timeline_create(
@@ -378,7 +597,10 @@ fn handle_tenant(tenant_match: &ArgMatches, env: &mut local_env::LocalEnv) -> an
                 Some(pg_version),
             )?;
             let new_timeline_id = timeline_info.timeline_id;
-            let last_record_lsn = timeline_info.last_record_lsn;
+            let last_record_lsn = timeline_info
+                .local
+                .context(format!("Failed to get last record LSN: no local timeline info for timeline {new_timeline_id}"))?
+                .last_record_lsn;
 
             env.register_branch_mapping(
                 DEFAULT_BRANCH_NAME.to_string(),
@@ -393,7 +615,7 @@ fn handle_tenant(tenant_match: &ArgMatches, env: &mut local_env::LocalEnv) -> an
         Some(("config", create_match)) => {
             let tenant_id = get_tenant_id(create_match, env)?;
             let tenant_conf: HashMap<_, _> = create_match
-                .get_many::<String>("config")
+                .values_of("config")
                 .map(|vals| vals.flat_map(|c| c.split_once(':')).collect())
                 .unwrap_or_default();
 
@@ -420,19 +642,23 @@ fn handle_timeline(timeline_match: &ArgMatches, env: &mut local_env::LocalEnv) -
         Some(("create", create_match)) => {
             let tenant_id = get_tenant_id(create_match, env)?;
             let new_branch_name = create_match
-                .get_one::<String>("branch-name")
+                .value_of("branch-name")
                 .ok_or_else(|| anyhow!("No branch name provided"))?;
 
             let pg_version = create_match
-                .get_one::<u32>("pg-version")
-                .copied()
+                .value_of("pg-version")
+                .unwrap()
+                .parse::<u32>()
                 .context("Failed to parse postgres version from the argument string")?;
 
             let timeline_info =
                 pageserver.timeline_create(tenant_id, None, None, None, Some(pg_version))?;
             let new_timeline_id = timeline_info.timeline_id;
 
-            let last_record_lsn = timeline_info.last_record_lsn;
+            let last_record_lsn = timeline_info
+                .local
+                .expect("no local timeline info")
+                .last_record_lsn;
             env.register_branch_mapping(new_branch_name.to_string(), tenant_id, new_timeline_id)?;
 
             println!(
@@ -444,32 +670,35 @@ fn handle_timeline(timeline_match: &ArgMatches, env: &mut local_env::LocalEnv) -
             let tenant_id = get_tenant_id(import_match, env)?;
             let timeline_id = parse_timeline_id(import_match)?.expect("No timeline id provided");
             let name = import_match
-                .get_one::<String>("node-name")
+                .value_of("node-name")
                 .ok_or_else(|| anyhow!("No node name provided"))?;
 
             // Parse base inputs
             let base_tarfile = import_match
-                .get_one::<PathBuf>("base-tarfile")
-                .ok_or_else(|| anyhow!("No base-tarfile provided"))?
-                .to_owned();
+                .value_of("base-tarfile")
+                .map(|s| PathBuf::from_str(s).unwrap())
+                .ok_or_else(|| anyhow!("No base-tarfile provided"))?;
             let base_lsn = Lsn::from_str(
                 import_match
-                    .get_one::<String>("base-lsn")
+                    .value_of("base-lsn")
                     .ok_or_else(|| anyhow!("No base-lsn provided"))?,
             )?;
             let base = (base_lsn, base_tarfile);
 
             // Parse pg_wal inputs
-            let wal_tarfile = import_match.get_one::<PathBuf>("wal-tarfile").cloned();
+            let wal_tarfile = import_match
+                .value_of("wal-tarfile")
+                .map(|s| PathBuf::from_str(s).unwrap());
             let end_lsn = import_match
-                .get_one::<String>("end-lsn")
+                .value_of("end-lsn")
                 .map(|s| Lsn::from_str(s).unwrap());
             // TODO validate both or none are provided
             let pg_wal = end_lsn.zip(wal_tarfile);
 
             let pg_version = import_match
-                .get_one::<u32>("pg-version")
-                .copied()
+                .value_of("pg-version")
+                .unwrap()
+                .parse::<u32>()
                 .context("Failed to parse postgres version from the argument string")?;
 
             let mut cplane = ComputeControlPlane::load(env.clone())?;
@@ -484,11 +713,10 @@ fn handle_timeline(timeline_match: &ArgMatches, env: &mut local_env::LocalEnv) -
         Some(("branch", branch_match)) => {
             let tenant_id = get_tenant_id(branch_match, env)?;
             let new_branch_name = branch_match
-                .get_one::<String>("branch-name")
+                .value_of("branch-name")
                 .ok_or_else(|| anyhow!("No branch name provided"))?;
             let ancestor_branch_name = branch_match
-                .get_one::<String>("ancestor-branch-name")
-                .map(|s| s.as_str())
+                .value_of("ancestor-branch-name")
                 .unwrap_or(DEFAULT_BRANCH_NAME);
             let ancestor_timeline_id = env
                 .get_branch_timeline_id(ancestor_branch_name, tenant_id)
@@ -497,8 +725,8 @@ fn handle_timeline(timeline_match: &ArgMatches, env: &mut local_env::LocalEnv) -
                 })?;
 
             let start_lsn = branch_match
-                .get_one::<String>("ancestor-start-lsn")
-                .map(|lsn_str| Lsn::from_str(lsn_str))
+                .value_of("ancestor-start-lsn")
+                .map(Lsn::from_str)
                 .transpose()
                 .context("Failed to parse ancestor start Lsn from the request")?;
             let timeline_info = pageserver.timeline_create(
@@ -510,7 +738,10 @@ fn handle_timeline(timeline_match: &ArgMatches, env: &mut local_env::LocalEnv) -
             )?;
             let new_timeline_id = timeline_info.timeline_id;
 
-            let last_record_lsn = timeline_info.last_record_lsn;
+            let last_record_lsn = timeline_info
+                .local
+                .expect("no local timeline info")
+                .last_record_lsn;
 
             env.register_branch_mapping(new_branch_name.to_string(), tenant_id, new_timeline_id)?;
 
@@ -570,7 +801,7 @@ fn handle_pg(pg_match: &ArgMatches, env: &local_env::LocalEnv) -> Result<()> {
                         // Use the LSN at the end of the timeline.
                         timeline_infos
                             .get(&node.timeline_id)
-                            .map(|bi| bi.last_record_lsn.to_string())
+                            .and_then(|bi| bi.local.as_ref().map(|l| l.last_record_lsn.to_string()))
                             .unwrap_or_else(|| "?".to_string())
                     }
                     Some(lsn) => {
@@ -599,39 +830,45 @@ fn handle_pg(pg_match: &ArgMatches, env: &local_env::LocalEnv) -> Result<()> {
         }
         "create" => {
             let branch_name = sub_args
-                .get_one::<String>("branch-name")
-                .map(|s| s.as_str())
+                .value_of("branch-name")
                 .unwrap_or(DEFAULT_BRANCH_NAME);
             let node_name = sub_args
-                .get_one::<String>("node")
-                .map(|node_name| node_name.to_string())
-                .unwrap_or_else(|| format!("{branch_name}_node"));
+                .value_of("node")
+                .map(ToString::to_string)
+                .unwrap_or_else(|| format!("{}_node", branch_name));
 
             let lsn = sub_args
-                .get_one::<String>("lsn")
-                .map(|lsn_str| Lsn::from_str(lsn_str))
+                .value_of("lsn")
+                .map(Lsn::from_str)
                 .transpose()
                 .context("Failed to parse Lsn from the request")?;
             let timeline_id = env
                 .get_branch_timeline_id(branch_name, tenant_id)
-                .ok_or_else(|| anyhow!("Found no timeline id for branch name '{branch_name}'"))?;
+                .ok_or_else(|| anyhow!("Found no timeline id for branch name '{}'", branch_name))?;
 
-            let port: Option<u16> = sub_args.get_one::<u16>("port").copied();
+            let port: Option<u16> = match sub_args.value_of("port") {
+                Some(p) => Some(p.parse()?),
+                None => None,
+            };
 
             let pg_version = sub_args
-                .get_one::<u32>("pg-version")
-                .copied()
+                .value_of("pg-version")
+                .unwrap()
+                .parse::<u32>()
                 .context("Failed to parse postgres version from the argument string")?;
 
             cplane.new_node(tenant_id, &node_name, timeline_id, lsn, port, pg_version)?;
         }
         "start" => {
-            let port: Option<u16> = sub_args.get_one::<u16>("port").copied();
+            let port: Option<u16> = match sub_args.value_of("port") {
+                Some(p) => Some(p.parse()?),
+                None => None,
+            };
             let node_name = sub_args
-                .get_one::<String>("node")
+                .value_of("node")
                 .ok_or_else(|| anyhow!("No node name was provided to start"))?;
 
-            let node = cplane.nodes.get(&(tenant_id, node_name.to_string()));
+            let node = cplane.nodes.get(&(tenant_id, node_name.to_owned()));
 
             let auth_token = if matches!(env.pageserver.auth_type, AuthType::NeonJWT) {
                 let claims = Claims::new(Some(tenant_id), Scope::Tenant);
@@ -642,33 +879,36 @@ fn handle_pg(pg_match: &ArgMatches, env: &local_env::LocalEnv) -> Result<()> {
             };
 
             if let Some(node) = node {
-                println!("Starting existing postgres {node_name}...");
+                println!("Starting existing postgres {}...", node_name);
                 node.start(&auth_token)?;
             } else {
                 let branch_name = sub_args
-                    .get_one::<String>("branch-name")
-                    .map(|s| s.as_str())
+                    .value_of("branch-name")
                     .unwrap_or(DEFAULT_BRANCH_NAME);
                 let timeline_id = env
                     .get_branch_timeline_id(branch_name, tenant_id)
                     .ok_or_else(|| {
-                        anyhow!("Found no timeline id for branch name '{branch_name}'")
+                        anyhow!("Found no timeline id for branch name '{}'", branch_name)
                     })?;
                 let lsn = sub_args
-                    .get_one::<String>("lsn")
-                    .map(|lsn_str| Lsn::from_str(lsn_str))
+                    .value_of("lsn")
+                    .map(Lsn::from_str)
                     .transpose()
                     .context("Failed to parse Lsn from the request")?;
                 let pg_version = sub_args
-                    .get_one::<u32>("pg-version")
-                    .copied()
-                    .context("Failed to `pg-version` from the argument string")?;
+                    .value_of("pg-version")
+                    .unwrap()
+                    .parse::<u32>()
+                    .context("Failed to parse postgres version from the argument string")?;
                 // when used with custom port this results in non obvious behaviour
                 // port is remembered from first start command, i e
                 // start --port X
                 // stop
                 // start <-- will also use port X even without explicit port argument
-                println!("Starting new postgres (v{pg_version}) {node_name} on timeline {timeline_id} ...");
+                println!(
+                    "Starting new postgres (v{}) {} on timeline {} ...",
+                    pg_version, node_name, timeline_id
+                );
 
                 let node =
                     cplane.new_node(tenant_id, node_name, timeline_id, lsn, port, pg_version)?;
@@ -677,18 +917,18 @@ fn handle_pg(pg_match: &ArgMatches, env: &local_env::LocalEnv) -> Result<()> {
         }
         "stop" => {
             let node_name = sub_args
-                .get_one::<String>("node")
+                .value_of("node")
                 .ok_or_else(|| anyhow!("No node name was provided to stop"))?;
-            let destroy = sub_args.get_flag("destroy");
+            let destroy = sub_args.is_present("destroy");
 
             let node = cplane
                 .nodes
-                .get(&(tenant_id, node_name.to_string()))
-                .with_context(|| format!("postgres {node_name} is not found"))?;
+                .get(&(tenant_id, node_name.to_owned()))
+                .with_context(|| format!("postgres {} is not found", node_name))?;
             node.stop(destroy)?;
         }
 
-        _ => bail!("Unexpected pg subcommand '{sub_name}'"),
+        _ => bail!("Unexpected pg subcommand '{}'", sub_name),
     }
 
     Ok(())
@@ -706,10 +946,7 @@ fn handle_pageserver(sub_match: &ArgMatches, env: &local_env::LocalEnv) -> Resul
         }
 
         Some(("stop", stop_match)) => {
-            let immediate = stop_match
-                .get_one::<String>("stop-mode")
-                .map(|s| s.as_str())
-                == Some("immediate");
+            let immediate = stop_match.value_of("stop-mode") == Some("immediate");
 
             if let Err(e) = pageserver.stop(immediate) {
                 eprintln!("pageserver stop failed: {}", e);
@@ -759,7 +996,7 @@ fn handle_safekeeper(sub_match: &ArgMatches, env: &local_env::LocalEnv) -> Resul
     };
 
     // All the commands take an optional safekeeper name argument
-    let sk_id = if let Some(id_str) = sub_args.get_one::<String>("id") {
+    let sk_id = if let Some(id_str) = sub_args.value_of("id") {
         NodeId(id_str.parse().context("while parsing safekeeper id")?)
     } else {
         DEFAULT_SAFEKEEPER_ID
@@ -775,8 +1012,7 @@ fn handle_safekeeper(sub_match: &ArgMatches, env: &local_env::LocalEnv) -> Resul
         }
 
         "stop" => {
-            let immediate =
-                sub_args.get_one::<String>("stop-mode").map(|s| s.as_str()) == Some("immediate");
+            let immediate = sub_args.value_of("stop-mode") == Some("immediate");
 
             if let Err(e) = safekeeper.stop(immediate) {
                 eprintln!("safekeeper stop failed: {}", e);
@@ -785,8 +1021,7 @@ fn handle_safekeeper(sub_match: &ArgMatches, env: &local_env::LocalEnv) -> Resul
         }
 
         "restart" => {
-            let immediate =
-                sub_args.get_one::<String>("stop-mode").map(|s| s.as_str()) == Some("immediate");
+            let immediate = sub_args.value_of("stop-mode") == Some("immediate");
 
             if let Err(e) = safekeeper.stop(immediate) {
                 eprintln!("safekeeper stop failed: {}", e);
@@ -830,8 +1065,7 @@ fn handle_start_all(sub_match: &ArgMatches, env: &local_env::LocalEnv) -> anyhow
 }
 
 fn handle_stop_all(sub_match: &ArgMatches, env: &local_env::LocalEnv) -> Result<()> {
-    let immediate =
-        sub_match.get_one::<String>("stop-mode").map(|s| s.as_str()) == Some("immediate");
+    let immediate = sub_match.value_of("stop-mode") == Some("immediate");
 
     let pageserver = PageServerNode::from_env(env);
 
@@ -864,219 +1098,3 @@ fn try_stop_etcd_process(env: &local_env::LocalEnv) {
         eprintln!("etcd stop failed: {e}");
     }
 }
-
-fn cli() -> Command {
-    let branch_name_arg = Arg::new("branch-name")
-        .long("branch-name")
-        .help("Name of the branch to be created or used as an alias for other services")
-        .required(false);
-
-    let pg_node_arg = Arg::new("node").help("Postgres node name").required(false);
-
-    let safekeeper_id_arg = Arg::new("id").help("safekeeper id").required(false);
-
-    let tenant_id_arg = Arg::new("tenant-id")
-        .long("tenant-id")
-        .help("Tenant id. Represented as a hexadecimal string 32 symbols length")
-        .required(false);
-
-    let timeline_id_arg = Arg::new("timeline-id")
-        .long("timeline-id")
-        .help("Timeline id. Represented as a hexadecimal string 32 symbols length")
-        .required(false);
-
-    let pg_version_arg = Arg::new("pg-version")
-        .long("pg-version")
-        .help("Postgres version to use for the initial tenant")
-        .required(false)
-        .value_parser(value_parser!(u32))
-        .default_value(DEFAULT_PG_VERSION);
-
-    let port_arg = Arg::new("port")
-        .long("port")
-        .required(false)
-        .value_parser(value_parser!(u16))
-        .value_name("port");
-
-    let stop_mode_arg = Arg::new("stop-mode")
-        .short('m')
-        .value_parser(["fast", "immediate"])
-        .help("If 'immediate', don't flush repository data at shutdown")
-        .required(false)
-        .value_name("stop-mode");
-
-    let pageserver_config_args = Arg::new("pageserver-config-override")
-        .long("pageserver-config-override")
-        .num_args(1)
-        .action(ArgAction::Append)
-        .help("Additional pageserver's configuration options or overrides, refer to pageserver's 'config-override' CLI parameter docs for more")
-        .required(false);
-
-    let lsn_arg = Arg::new("lsn")
-        .long("lsn")
-        .help("Specify Lsn on the timeline to start from. By default, end of the timeline would be used.")
-        .required(false);
-
-    Command::new("Neon CLI")
-        .arg_required_else_help(true)
-        .version(GIT_VERSION)
-        .subcommand(
-            Command::new("init")
-                .about("Initialize a new Neon repository")
-                .arg(pageserver_config_args.clone())
-                .arg(timeline_id_arg.clone().help("Use a specific timeline id when creating a tenant and its initial timeline"))
-                .arg(
-                    Arg::new("config")
-                        .long("config")
-                        .required(false)
-                        .value_parser(value_parser!(PathBuf))
-                        .value_name("config"),
-                )
-                .arg(pg_version_arg.clone())
-        )
-        .subcommand(
-            Command::new("timeline")
-            .about("Manage timelines")
-            .subcommand(Command::new("list")
-                .about("List all timelines, available to this pageserver")
-                .arg(tenant_id_arg.clone()))
-            .subcommand(Command::new("branch")
-                .about("Create a new timeline, using another timeline as a base, copying its data")
-                .arg(tenant_id_arg.clone())
-                .arg(branch_name_arg.clone())
-                .arg(Arg::new("ancestor-branch-name").long("ancestor-branch-name")
-                    .help("Use last Lsn of another timeline (and its data) as base when creating the new timeline. The timeline gets resolved by its branch name.").required(false))
-                .arg(Arg::new("ancestor-start-lsn").long("ancestor-start-lsn")
-                    .help("When using another timeline as base, use a specific Lsn in it instead of the latest one").required(false)))
-            .subcommand(Command::new("create")
-                .about("Create a new blank timeline")
-                .arg(tenant_id_arg.clone())
-                .arg(branch_name_arg.clone())
-                .arg(pg_version_arg.clone())
-            )
-            .subcommand(Command::new("import")
-                .about("Import timeline from basebackup directory")
-                .arg(tenant_id_arg.clone())
-                .arg(timeline_id_arg.clone())
-                .arg(Arg::new("node-name").long("node-name")
-                    .help("Name to assign to the imported timeline"))
-                .arg(Arg::new("base-tarfile")
-                    .long("base-tarfile")
-                    .value_parser(value_parser!(PathBuf))
-                    .help("Basebackup tarfile to import")
-                )
-                .arg(Arg::new("base-lsn").long("base-lsn")
-                    .help("Lsn the basebackup starts at"))
-                .arg(Arg::new("wal-tarfile")
-                    .long("wal-tarfile")
-                    .value_parser(value_parser!(PathBuf))
-                    .help("Wal to add after base")
-                )
-                .arg(Arg::new("end-lsn").long("end-lsn")
-                    .help("Lsn the basebackup ends at"))
-                .arg(pg_version_arg.clone())
-            )
-        ).subcommand(
-            Command::new("tenant")
-            .arg_required_else_help(true)
-            .about("Manage tenants")
-            .subcommand(Command::new("list"))
-            .subcommand(Command::new("create")
-                .arg(tenant_id_arg.clone())
-                .arg(timeline_id_arg.clone().help("Use a specific timeline id when creating a tenant and its initial timeline"))
-                .arg(Arg::new("config").short('c').num_args(1).action(ArgAction::Append).required(false))
-                .arg(pg_version_arg.clone())
-                )
-            .subcommand(Command::new("config")
-                .arg(tenant_id_arg.clone())
-                .arg(Arg::new("config").short('c').num_args(1).action(ArgAction::Append).required(false))
-                )
-        )
-        .subcommand(
-            Command::new("pageserver")
-                .arg_required_else_help(true)
-                .about("Manage pageserver")
-                .subcommand(Command::new("status"))
-                .subcommand(Command::new("start").about("Start local pageserver").arg(pageserver_config_args.clone()))
-                .subcommand(Command::new("stop").about("Stop local pageserver")
-                            .arg(stop_mode_arg.clone()))
-                .subcommand(Command::new("restart").about("Restart local pageserver").arg(pageserver_config_args.clone()))
-        )
-        .subcommand(
-            Command::new("safekeeper")
-                .arg_required_else_help(true)
-                .about("Manage safekeepers")
-                .subcommand(Command::new("start")
-                            .about("Start local safekeeper")
-                            .arg(safekeeper_id_arg.clone())
-                )
-                .subcommand(Command::new("stop")
-                            .about("Stop local safekeeper")
-                            .arg(safekeeper_id_arg.clone())
-                            .arg(stop_mode_arg.clone())
-                )
-                .subcommand(Command::new("restart")
-                            .about("Restart local safekeeper")
-                            .arg(safekeeper_id_arg)
-                            .arg(stop_mode_arg.clone())
-                )
-        )
-        .subcommand(
-            Command::new("pg")
-                .arg_required_else_help(true)
-                .about("Manage postgres instances")
-                .subcommand(Command::new("list").arg(tenant_id_arg.clone()))
-                .subcommand(Command::new("create")
-                    .about("Create a postgres compute node")
-                    .arg(pg_node_arg.clone())
-                    .arg(branch_name_arg.clone())
-                    .arg(tenant_id_arg.clone())
-                    .arg(lsn_arg.clone())
-                    .arg(port_arg.clone())
-                    .arg(
-                        Arg::new("config-only")
-                            .help("Don't do basebackup, create compute node with only config files")
-                            .long("config-only")
-                            .required(false))
-                    .arg(pg_version_arg.clone())
-                )
-                .subcommand(Command::new("start")
-                    .about("Start a postgres compute node.\n This command actually creates new node from scratch, but preserves existing config files")
-                    .arg(pg_node_arg.clone())
-                    .arg(tenant_id_arg.clone())
-                    .arg(branch_name_arg)
-                    .arg(timeline_id_arg)
-                    .arg(lsn_arg)
-                    .arg(port_arg)
-                    .arg(pg_version_arg)
-                )
-                .subcommand(
-                    Command::new("stop")
-                    .arg(pg_node_arg)
-                    .arg(tenant_id_arg)
-                    .arg(
-                        Arg::new("destroy")
-                            .help("Also delete data directory (now optional, should be default in future)")
-                            .long("destroy")
-                            .action(ArgAction::SetTrue)
-                            .required(false)
-                        )
-                )
-
-        )
-        .subcommand(
-            Command::new("start")
-                .about("Start page server and safekeepers")
-                .arg(pageserver_config_args)
-        )
-        .subcommand(
-            Command::new("stop")
-                .about("Stop page server and safekeepers")
-                .arg(stop_mode_arg)
-        )
-}
-
-#[test]
-fn verify_cli() {
-    cli().debug_assert();
-}

control_plane/src/compute.rs

@@ -183,18 +183,18 @@ impl PostgresNode {
     }
 
     fn sync_safekeepers(&self, auth_token: &Option<String>, pg_version: u32) -> Result<Lsn> {
-        let pg_path = self.env.pg_bin_dir(pg_version)?.join("postgres");
+        let pg_path = self.env.pg_bin_dir(pg_version).join("postgres");
         let mut cmd = Command::new(&pg_path);
 
         cmd.arg("--sync-safekeepers")
             .env_clear()
             .env(
                 "LD_LIBRARY_PATH",
-                self.env.pg_lib_dir(pg_version)?.to_str().unwrap(),
+                self.env.pg_lib_dir(pg_version).to_str().unwrap(),
             )
             .env(
                 "DYLD_LIBRARY_PATH",
-                self.env.pg_lib_dir(pg_version)?.to_str().unwrap(),
+                self.env.pg_lib_dir(pg_version).to_str().unwrap(),
             )
             .env("PGDATA", self.pgdata().to_str().unwrap())
             .stdout(Stdio::piped())
@@ -282,7 +282,9 @@ impl PostgresNode {
     fn setup_pg_conf(&self, auth_type: AuthType) -> Result<()> {
         let mut conf = PostgresConf::new();
         conf.append("max_wal_senders", "10");
-        conf.append("wal_log_hints", "off");
+        // wal_log_hints is mandatory when running against pageserver (see gh issue#192)
+        // TODO: is it possible to check wal_log_hints at pageserver side via XLOG_PARAMETER_CHANGE?
+        conf.append("wal_log_hints", "on");
         conf.append("max_replication_slots", "10");
         conf.append("hot_standby", "on");
         conf.append("shared_buffers", "1MB");
@@ -420,7 +422,7 @@ impl PostgresNode {
     }
 
     fn pg_ctl(&self, args: &[&str], auth_token: &Option<String>) -> Result<()> {
-        let pg_ctl_path = self.env.pg_bin_dir(self.pg_version)?.join("pg_ctl");
+        let pg_ctl_path = self.env.pg_bin_dir(self.pg_version).join("pg_ctl");
         let mut cmd = Command::new(pg_ctl_path);
         cmd.args(
             [
@@ -438,11 +440,11 @@ impl PostgresNode {
         .env_clear()
         .env(
             "LD_LIBRARY_PATH",
-            self.env.pg_lib_dir(self.pg_version)?.to_str().unwrap(),
+            self.env.pg_lib_dir(self.pg_version).to_str().unwrap(),
         )
         .env(
             "DYLD_LIBRARY_PATH",
-            self.env.pg_lib_dir(self.pg_version)?.to_str().unwrap(),
+            self.env.pg_lib_dir(self.pg_version).to_str().unwrap(),
         );
         if let Some(token) = auth_token {
             cmd.env("ZENITH_AUTH_TOKEN", token);

control_plane/src/etcd.rs

@@ -52,10 +52,6 @@ pub fn start_etcd_process(env: &local_env::LocalEnv) -> anyhow::Result<()> {
             // size smaller. Our test etcd clusters are very small.
             // See https://github.com/etcd-io/etcd/issues/7910
             "--quota-backend-bytes=100000000".to_string(),
-            // etcd doesn't compact (vacuum) with default settings,
-            // enable it to prevent space exhaustion.
-            "--auto-compaction-mode=revision".to_string(),
-            "--auto-compaction-retention=1".to_string(),
         ])
         .stdout(Stdio::from(etcd_stdout_file))
         .stderr(Stdio::from(etcd_stderr_file))

control_plane/src/local_env.rs

@@ -201,28 +201,28 @@ impl LocalEnv {
         self.pg_distrib_dir.clone()
     }
 
-    pub fn pg_distrib_dir(&self, pg_version: u32) -> anyhow::Result<PathBuf> {
+    pub fn pg_distrib_dir(&self, pg_version: u32) -> PathBuf {
         let path = self.pg_distrib_dir.clone();
 
         match pg_version {
-            14 => Ok(path.join(format!("v{pg_version}"))),
-            15 => Ok(path.join(format!("v{pg_version}"))),
-            _ => bail!("Unsupported postgres version: {}", pg_version),
+            14 => path.join(format!("v{pg_version}")),
+            15 => path.join(format!("v{pg_version}")),
+            _ => panic!("Unsupported postgres version: {}", pg_version),
         }
     }
 
-    pub fn pg_bin_dir(&self, pg_version: u32) -> anyhow::Result<PathBuf> {
+    pub fn pg_bin_dir(&self, pg_version: u32) -> PathBuf {
         match pg_version {
-            14 => Ok(self.pg_distrib_dir(pg_version)?.join("bin")),
-            15 => Ok(self.pg_distrib_dir(pg_version)?.join("bin")),
-            _ => bail!("Unsupported postgres version: {}", pg_version),
+            14 => self.pg_distrib_dir(pg_version).join("bin"),
+            15 => self.pg_distrib_dir(pg_version).join("bin"),
+            _ => panic!("Unsupported postgres version: {}", pg_version),
         }
     }
-    pub fn pg_lib_dir(&self, pg_version: u32) -> anyhow::Result<PathBuf> {
+    pub fn pg_lib_dir(&self, pg_version: u32) -> PathBuf {
         match pg_version {
-            14 => Ok(self.pg_distrib_dir(pg_version)?.join("lib")),
-            15 => Ok(self.pg_distrib_dir(pg_version)?.join("lib")),
-            _ => bail!("Unsupported postgres version: {}", pg_version),
+            14 => self.pg_distrib_dir(pg_version).join("lib"),
+            15 => self.pg_distrib_dir(pg_version).join("lib"),
+            _ => panic!("Unsupported postgres version: {}", pg_version),
         }
     }
 
@@ -422,10 +422,10 @@ impl LocalEnv {
             "directory '{}' already exists. Perhaps already initialized?",
             base_path.display()
         );
-        if !self.pg_bin_dir(pg_version)?.join("postgres").exists() {
+        if !self.pg_bin_dir(pg_version).join("postgres").exists() {
             bail!(
                 "Can't find postgres binary at {}",
-                self.pg_bin_dir(pg_version)?.display()
+                self.pg_bin_dir(pg_version).display()
             );
         }
         for binary in ["pageserver", "safekeeper"] {

control_plane/src/safekeeper.rs

@@ -12,8 +12,13 @@ use nix::unistd::Pid;
 use postgres::Config;
 use reqwest::blocking::{Client, RequestBuilder, Response};
 use reqwest::{IntoUrl, Method};
+use safekeeper_api::models::TimelineCreateRequest;
 use thiserror::Error;
-use utils::{connstring::connection_address, http::error::HttpErrorBody, id::NodeId};
+use utils::{
+    connstring::connection_address,
+    http::error::HttpErrorBody,
+    id::{NodeId, TenantId, TimelineId},
+};
 
 use crate::local_env::{LocalEnv, SafekeeperConf};
 use crate::storage::PageServerNode;
@@ -123,6 +128,7 @@ impl SafekeeperNode {
                 .args(&["--id", self.id.to_string().as_ref()])
                 .args(&["--listen-pg", &listen_pg])
                 .args(&["--listen-http", &listen_http])
+                .args(&["--recall", "1 second"])
                 .arg("--daemonize"),
         );
         if !self.conf.sync {
@@ -275,4 +281,24 @@ impl SafekeeperNode {
             .error_from_body()?;
         Ok(())
     }
+
+    pub fn timeline_create(
+        &self,
+        tenant_id: TenantId,
+        timeline_id: TimelineId,
+        peer_ids: Vec<NodeId>,
+    ) -> Result<()> {
+        Ok(self
+            .http_request(
+                Method::POST,
+                format!("{}/tenant/{}/timeline", self.http_base_url, tenant_id),
+            )
+            .json(&TimelineCreateRequest {
+                timeline_id,
+                peer_ids,
+            })
+            .send()?
+            .error_from_body()?
+            .json()?)
+    }
 }

docker-compose/compute/shell/compute.sh

@@ -1,48 +0,0 @@
-#!/bin/bash
-set -eux
-
-PG_VERSION=${PG_VERSION:-14}
-
-SPEC_FILE_ORG=/var/db/postgres/specs/spec.json
-SPEC_FILE=/tmp/spec.json
-
-echo "Waiting pageserver become ready."
-while ! nc -z pageserver 6400; do
-     sleep 1;
-done
-echo "Page server is ready."
-
-echo "Create a tenant and timeline"
-PARAMS=(
-     -sb 
-     -X POST
-     -H "Content-Type: application/json"
-     -d "{}"
-     http://pageserver:9898/v1/tenant/
-)
-tenant_id=$(curl "${PARAMS[@]}" | sed 's/"//g')
-
-PARAMS=(
-     -sb 
-     -X POST
-     -H "Content-Type: application/json"
-     -d "{\"tenant_id\":\"${tenant_id}\", \"pg_version\": ${PG_VERSION}}"
-     "http://pageserver:9898/v1/tenant/${tenant_id}/timeline/"
-)
-result=$(curl "${PARAMS[@]}")
-echo $result | jq .
-
-echo "Overwrite tenant id and timeline id in spec file"
-tenant_id=$(echo ${result} | jq -r .tenant_id)
-timeline_id=$(echo ${result} | jq -r .timeline_id)
-
-sed "s/TENANT_ID/${tenant_id}/" ${SPEC_FILE_ORG} > ${SPEC_FILE}
-sed -i "s/TIMELINE_ID/${timeline_id}/" ${SPEC_FILE}
-
-cat ${SPEC_FILE}
-
-echo "Start compute node"
-/usr/local/bin/compute_ctl --pgdata /var/db/postgres/compute \
-     -C "postgresql://cloud_admin@localhost:55433/postgres"  \
-     -b /usr/local/bin/postgres                              \
-     -S ${SPEC_FILE}

docker-compose/compute/var/db/postgres/specs/spec.json

@@ -1,141 +0,0 @@
-{
-    "format_version": 1.0,
-
-    "timestamp": "2022-10-12T18:00:00.000Z",
-    "operation_uuid": "0f657b36-4b0f-4a2d-9c2e-1dcd615e7d8c",
-
-    "cluster": {
-        "cluster_id": "docker_compose",
-        "name": "docker_compose_test",
-        "state": "restarted",
-        "roles": [
-            {
-                "name": "cloud_admin",
-                "encrypted_password": "b093c0d3b281ba6da1eacc608620abd8",
-                "options": null
-            }
-        ],
-        "databases": [
-        ],
-        "settings": [
-            {
-                "name": "fsync",
-                "value": "off",
-                "vartype": "bool"
-            },
-            {
-                "name": "wal_level",
-                "value": "replica",
-                "vartype": "enum"
-            },
-            {
-                "name": "hot_standby",
-                "value": "on",
-                "vartype": "bool"
-            },
-            {
-                "name": "wal_log_hints",
-                "value": "on",
-                "vartype": "bool"
-            },
-            {
-                "name": "log_connections",
-                "value": "on",
-                "vartype": "bool"
-            },
-            {
-                "name": "port",
-                "value": "55433",
-                "vartype": "integer"
-            },
-            {
-                "name": "shared_buffers",
-                "value": "1MB",
-                "vartype": "string"
-            },
-            {
-                "name": "max_connections",
-                "value": "100",
-                "vartype": "integer"
-            },
-            {
-                "name": "listen_addresses",
-                "value": "0.0.0.0",
-                "vartype": "string"
-            },
-            {
-                "name": "max_wal_senders",
-                "value": "10",
-                "vartype": "integer"
-            },
-            {
-                "name": "max_replication_slots",
-                "value": "10",
-                "vartype": "integer"
-            },
-            {
-                "name": "wal_sender_timeout",
-                "value": "5s",
-                "vartype": "string"
-            },
-            {
-                "name": "wal_keep_size",
-                "value": "0",
-                "vartype": "integer"
-            },
-            {
-                "name": "password_encryption",
-                "value": "md5",
-                "vartype": "enum"
-            },
-            {
-                "name": "restart_after_crash",
-                "value": "off",
-                "vartype": "bool"
-            },
-            {
-                "name": "synchronous_standby_names",
-                "value": "walproposer",
-                "vartype": "string"
-            },
-            {
-                "name": "shared_preload_libraries",
-                "value": "neon",
-                "vartype": "string"
-            },
-            {
-                "name": "neon.safekeepers",
-                "value": "safekeeper1:5454,safekeeper2:5454,safekeeper3:5454",
-                "vartype": "string"
-            },
-            {
-                "name": "neon.timeline_id",
-                "value": "TIMELINE_ID",
-                "vartype": "string"
-            },
-            {
-                "name": "neon.tenant_id",
-                "value": "TENANT_ID",
-                "vartype": "string"
-            },
-            {
-                "name": "neon.pageserver_connstring",
-                "value": "host=pageserver port=6400",
-                "vartype": "string"
-            },
-            {
-                "name": "max_replication_write_lag",
-                "value": "500MB",
-                "vartype": "string"
-            },
-            {
-                "name": "max_replication_flush_lag",
-                "value": "10GB",
-                "vartype": "string"
-            }
-        ]
-    },
-
-    "delta_operations": [
-    ]
-}

docker-compose/docker-compose.yml

@@ -1,200 +0,0 @@
-version: '3'
-
-services:
-  etcd:
-    image: quay.io/coreos/etcd:v3.5.4
-    ports:
-      - 2379:2379
-      - 2380:2380
-    environment:
-      # This signifficantly speeds up etcd and we anyway don't data persistency there.
-      ETCD_UNSAFE_NO_FSYNC: "1"
-    command: 
-      - "etcd"
-      - "--auto-compaction-mode=revision"
-      - "--auto-compaction-retention=1"
-      - "--name=etcd-cluster"
-      - "--initial-cluster-state=new"
-      - "--initial-cluster-token=etcd-cluster-1"
-      - "--initial-cluster=etcd-cluster=http://etcd:2380"
-      - "--initial-advertise-peer-urls=http://etcd:2380"
-      - "--advertise-client-urls=http://etcd:2379"
-      - "--listen-client-urls=http://0.0.0.0:2379"
-      - "--listen-peer-urls=http://0.0.0.0:2380"
-      - "--quota-backend-bytes=134217728" # 128 MB
-
-  minio:
-    image: quay.io/minio/minio:RELEASE.2022-10-20T00-55-09Z
-    ports:
-      - 9000:9000
-      - 9001:9001
-    environment:
-      - MINIO_ROOT_USER=minio
-      - MINIO_ROOT_PASSWORD=password
-    command: server /data --address :9000 --console-address ":9001"
-
-  minio_create_buckets:
-    image: minio/mc
-    environment:
-      - MINIO_ROOT_USER=minio
-      - MINIO_ROOT_PASSWORD=password
-    entrypoint:
-      - "/bin/sh"
-      - "-c"
-    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;
-         /usr/bin/mc mb minio/neon --region=eu-north-1;
-         exit 0;"
-    depends_on:
-      - minio
-
-  pageserver:
-    image: neondatabase/neon:${TAG:-latest}
-    environment:
-      - BROKER_ENDPOINT='http://etcd:2379'
-      - AWS_ACCESS_KEY_ID=minio
-      - AWS_SECRET_ACCESS_KEY=password
-      #- RUST_BACKTRACE=1
-    ports:
-       #- 6400:6400  # pg protocol handler
-       - 9898:9898 # http endpoints
-    entrypoint:
-      - "/bin/sh"
-      - "-c"
-    command:
-      - "/usr/local/bin/pageserver -D /data/.neon/
-                                   -c \"broker_endpoints=[$$BROKER_ENDPOINT]\"
-                                   -c \"listen_pg_addr='0.0.0.0:6400'\"
-                                   -c \"listen_http_addr='0.0.0.0:9898'\"
-                                   -c \"remote_storage={endpoint='http://minio:9000',
-                                                        bucket_name='neon',
-                                                        bucket_region='eu-north-1',
-                                                        prefix_in_bucket='/pageserver/'}\""
-    depends_on:
-      - etcd
-      - minio_create_buckets
-
-  safekeeper1:
-    image: neondatabase/neon:${TAG:-latest}
-    environment:
-      - SAFEKEEPER_ADVERTISE_URL=safekeeper1:5454
-      - SAFEKEEPER_ID=1
-      - BROKER_ENDPOINT=http://etcd:2379
-      - AWS_ACCESS_KEY_ID=minio
-      - AWS_SECRET_ACCESS_KEY=password
-      #- RUST_BACKTRACE=1
-    ports:
-      #- 5454:5454 # pg protocol handler
-      - 7676:7676 # http endpoints
-    entrypoint:
-      - "/bin/sh"
-      - "-c"
-    command:
-      - "safekeeper --listen-pg=$$SAFEKEEPER_ADVERTISE_URL
-                    --listen-http='0.0.0.0:7676'
-                    --id=$$SAFEKEEPER_ID
-                    --broker-endpoints=$$BROKER_ENDPOINT
-                    -D /data
-                    --remote-storage=\"{endpoint='http://minio:9000',
-                                        bucket_name='neon',
-                                        bucket_region='eu-north-1',
-                                        prefix_in_bucket='/safekeeper/'}\""
-    depends_on:
-      - etcd
-      - minio_create_buckets
-
-  safekeeper2:
-    image: neondatabase/neon:${TAG:-latest}
-    environment:
-      - SAFEKEEPER_ADVERTISE_URL=safekeeper2:5454
-      - SAFEKEEPER_ID=2
-      - BROKER_ENDPOINT=http://etcd:2379
-      - AWS_ACCESS_KEY_ID=minio
-      - AWS_SECRET_ACCESS_KEY=password
-      #- RUST_BACKTRACE=1
-    ports:
-      #- 5454:5454 # pg protocol handler
-      - 7677:7676 # http endpoints
-    entrypoint:
-      - "/bin/sh"
-      - "-c"
-    command:
-      - "safekeeper --listen-pg=$$SAFEKEEPER_ADVERTISE_URL
-                    --listen-http='0.0.0.0:7676'
-                    --id=$$SAFEKEEPER_ID
-                    --broker-endpoints=$$BROKER_ENDPOINT
-                    -D /data
-                    --remote-storage=\"{endpoint='http://minio:9000',
-                                        bucket_name='neon',
-                                        bucket_region='eu-north-1',
-                                        prefix_in_bucket='/safekeeper/'}\""
-    depends_on:
-      - etcd
-      - minio_create_buckets
-
-  safekeeper3:
-    image: neondatabase/neon:${TAG:-latest}
-    environment:
-      - SAFEKEEPER_ADVERTISE_URL=safekeeper3:5454
-      - SAFEKEEPER_ID=3
-      - BROKER_ENDPOINT=http://etcd:2379
-      - AWS_ACCESS_KEY_ID=minio
-      - AWS_SECRET_ACCESS_KEY=password
-      #- RUST_BACKTRACE=1
-    ports:
-      #- 5454:5454 # pg protocol handler
-      - 7678:7676 # http endpoints
-    entrypoint:
-      - "/bin/sh"
-      - "-c"
-    command:
-      - "safekeeper --listen-pg=$$SAFEKEEPER_ADVERTISE_URL
-                    --listen-http='0.0.0.0:7676'
-                    --id=$$SAFEKEEPER_ID
-                    --broker-endpoints=$$BROKER_ENDPOINT
-                    -D /data
-                    --remote-storage=\"{endpoint='http://minio:9000',
-                                        bucket_name='neon',
-                                        bucket_region='eu-north-1',
-                                        prefix_in_bucket='/safekeeper/'}\""
-    depends_on:
-      - etcd
-      - minio_create_buckets
-
-  compute:
-    build:
-      context: ./image/compute
-      args:
-        - COMPUTE_IMAGE=compute-node-v${PG_VERSION:-14}:${TAG:-latest}
-        - http_proxy=$http_proxy
-        - https_proxy=$https_proxy
-    environment:
-      - PG_VERSION=${PG_VERSION:-14}
-      #- RUST_BACKTRACE=1
-    volumes:
-      - ./compute/var/db/postgres/specs/:/var/db/postgres/specs/
-      - ./compute/shell/:/shell/
-    ports:
-      - 55433:55433 # pg protocol handler
-      - 3080:3080 # http endpoints
-    entrypoint:
-      - "/shell/compute.sh"
-    depends_on:
-      - safekeeper1
-      - safekeeper2
-      - safekeeper3
-      - pageserver
-
-  compute_is_ready:
-    image: postgres:latest
-    entrypoint:
-      - "/bin/bash"
-      - "-c"
-    command:
-      - "until pg_isready -h compute -p 55433 ; do
-            echo 'Waiting to start compute...' && sleep 1;
-         done"
-    depends_on:
-      - compute

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

@@ -80,6 +80,4 @@
 - [015-storage-messaging](rfcs/015-storage-messaging.md)
 - [016-connection-routing](rfcs/016-connection-routing.md)
 - [017-timeline-data-management](rfcs/017-timeline-data-management.md)
-- [018-storage-messaging-2](rfcs/018-storage-messaging-2.md)
-- [019-tenant-timeline-lifecycles](rfcs/019-tenant-timeline-lifecycles.md)
 - [cluster-size-limits](rfcs/cluster-size-limits.md)

docs/docker.md

@@ -18,67 +18,3 @@ We build all images after a successful `release` tests run and push automaticall
 1. `neondatabase/compute-tools` and `neondatabase/compute-node`
 
 2. `neondatabase/neon`
-
-## Docker Compose example
-
-You can see a [docker compose](https://docs.docker.com/compose/) example to create a neon cluster in [/docker-compose/docker-compose.yml](/docker-compose/docker-compose.yml). It creates the following conatainers.
-
-- etcd x 1
-- pageserver x 1
-- safekeeper x 3
-- compute x 1
-- MinIO x 1        # This is Amazon S3 compatible object storage
-
-### How to use
-
-1. create containers
-
-You can specify version of neon cluster using following environment values.
-- PG_VERSION: postgres version for compute (default is 14)
-- TAG: the tag version of [docker image](https://registry.hub.docker.com/r/neondatabase/neon/tags) (default is latest), which is tagged in [CI test](/.github/workflows/build_and_test.yml)
-```
-$ cd docker-compose/docker-compose.yml
-$ docker-compose down   # remove the conainers if exists
-$ PG_VERSION=15 TAG=2221 docker-compose up --build -d  # You can specify the postgres and image version
-Creating network "dockercompose_default" with the default driver
-Creating dockercompose_etcd3_1 ...
-(...omit...)
-```
-
-2. connect compute node
-```
-$ echo "localhost:55433:postgres:cloud_admin:cloud_admin" >> ~/.pgpass
-$ psql -h localhost -p 55433 -U cloud_admin
-postgres=# CREATE TABLE t(key int primary key, value text);
-CREATE TABLE
-postgres=# insert into t values(1,1);
-INSERT 0 1
-postgres=# select * from t;
- key | value
------+-------
-   1 | 1
-(1 row)
-```
-
-3. If you want to see the log, you can use `docker-compose logs` command.
-```
-# check the container name you want to see
-$ docker ps
-CONTAINER ID   IMAGE                                              COMMAND                  CREATED         STATUS         PORTS                                                                                                                                  NAMES
-d6968a5ae912   dockercompose_compute                              "/shell/compute.sh"      5 minutes ago   Up 5 minutes   0.0.0.0:3080->3080/tcp, 0.0.0.0:55433->55433/tcp                                                                                       dockercompose_compute_1
-(...omit...)
-
-$ docker logs -f dockercompose_compute_1
-2022-10-21 06:15:48.757 GMT [56] LOG:  connection authorized: user=cloud_admin database=postgres application_name=psql
-2022-10-21 06:17:00.307 GMT [56] LOG:  [NEON_SMGR] libpagestore: connected to 'host=pageserver port=6400'
-(...omit...)
-```
-
-4. If you want to see durable data in MinIO which is s3 compatible storage
-
-Access http://localhost:9001 and sign in.
-
-- Username: `minio`
-- Password: `password`
-
-You can see durable pages and WAL data in `neon` bucket.

docs/rfcs/019-tenant-timeline-lifecycles.md

@@ -1,91 +0,0 @@
-# Managing Tenant and Timeline lifecycles
-
-## Summary
-
-The pageserver has a Tenant object in memory for each tenant it manages, and a
-Timeline for each timeline. There are a lot of tasks that operate on the tenants
-and timelines with references to those objects. We have some mechanisms to track
-which tasks are operating on each Tenant and Timeline, and to request them to
-shutdown when a tenant or timeline is deleted, but it does not cover all uses,
-and as a result we have many race conditions around tenant/timeline shutdown.
-
-## Motivation
-
-We have a bunch of race conditions that can produce weird errors and can be hard
-to track down.
-
-## Non Goals
-
-This RFC only covers the problem of ensuring that a task/thread isn't operating
-on a Tenant or Timeline. It does not cover what states, aside from Active and
-non-Active, each Tenant and Timeline should have, or when exactly the transitions
-should happen.
-
-## Impacted components (e.g. pageserver, safekeeper, console, etc)
-
-Pageserver. Although I wonder if the safekeeper should have a similar mechanism.
-
-## Current situation
-
-Most pageserver tasks of are managed by task_mgr.rs:
-
-- LibpqEndpointListener
-- HttpEndPointListener
-- WalReceiverManager and -Connection
-- GarbageCollector and Compaction
-- InitialLogicalSizeCalculation
-
-In addition to those tasks, the walreceiver performs some direct tokio::spawn
-calls to spawn tasks that are not registered with 'task_mgr'. And all of these
-tasks can spawn extra operations with tokio spawn_blocking.
-
-Whenever a tenant or timeline is removed from the system, by pageserver
-shutdown, delete_timeline or tenant-detach operation, we rely on the task
-registry in 'task_mgr.rs' to wait until there are no tasks operating on the
-tenant or timeline, before its Tenant/Timeline object is removed. That relies on
-each task to register itself with the tenant/timeline ID in
-'task_mgr.rs'. However, there are many gaps in that. For example,
-GarbageCollection and Compaction tasks are registered with the tenant, but when
-they proceed to operate on a particular timeline of the tenant, they don't
-register with timeline ID. Because of that, the timeline can be deleted while GC
-or compaction is running on it, causing failures in the GC or compaction (see
-https://github.com/neondatabase/neon/issues/2442).
-
-Another problem is that the task registry only works for tokio Tasks. There is
-no way to register a piece of code that runs inside spawn_blocking(), for
-example.
-
-## Proposed implementation
-
-This "voluntary" registration of tasks is fragile. Let's use Rust language features
-to enforce that a tenant/timeline cannot be removed from the system when there is
-still some code operating on it.
-
-Let's introduce new Guard objects for Tenant and Timeline, and do all actions through
-the Guard object. Something like:
-
-TenantActiveGuard: Guard object over Arc<Tenant>. When you acquire the guard,
-the code checks that the tenant is in Active state. If it's not, you get an
-error. You can change the state of the tenant to Stopping while there are
-ActiveTenantGuard objects still on it, to prevent new ActiveTenantGuards from
-being acquired, but the Tenant cannot be removed until all the guards are gone.
-
-TenantMaintenanceGuard: Like ActiveTenantGuard, but can be held even when the
-tenant is not in Active state. Used for operations like attach/detach. Perhaps
-allow only one such guard on a Tenant at a time.
-
-Similarly for Timelines. We don't currentl have a "state" on Timeline, but I think
-we need at least two states: Active and Stopping. The Stopping state is used at
-deletion, to prevent new TimelineActiveGuards from appearing, while you wait for
-existing TimelineActiveGuards to die out.
-
-The shutdown-signaling, using shutdown_watcher() and is_shutdown_requested(),
-probably also needs changes to deal with the new Guards. The rule is that if you
-have a TenantActiveGuard, and the tenant's state changes from Active to
-Stopping, the is_shutdown_requested() function should return true, and
-shutdown_watcher() future should return.
-
-This signaling doesn't neessarily need to cover all cases. For example, if you
-have a block of code in spawn_blocking(), it might be acceptable if
-is_shutdown_requested() doesn't return true even though the tenant is in
-Stopping state, as long as the code finishes reasonably fast.

libs/etcd_broker/Cargo.toml

@@ -8,7 +8,7 @@
  regex = "1.4.5"
  serde = { version = "1.0", features = ["derive"] }
  serde_json = "1"
- serde_with = "2.0"
+ serde_with = "1.12.0"
  once_cell = "1.13.0"
 
  utils = { path = "../utils" }

libs/etcd_broker/src/subscription_value.rs

@@ -29,9 +29,6 @@ pub struct SkTimelineInfo {
     #[serde_as(as = "Option<DisplayFromStr>")]
     #[serde(default)]
     pub peer_horizon_lsn: Option<Lsn>,
-    #[serde_as(as = "Option<DisplayFromStr>")]
-    #[serde(default)]
-    pub local_start_lsn: Option<Lsn>,
     /// A connection string to use for WAL receiving.
     #[serde(default)]
     pub safekeeper_connstr: Option<String>,

libs/pageserver_api/Cargo.toml

@@ -5,11 +5,8 @@ edition = "2021"
 
 [dependencies]
 serde = { version = "1.0", features = ["derive"] }
-serde_with = "2.0"
+serde_with = "1.12.0"
 const_format = "0.2.21"
-anyhow = { version = "1.0", features = ["backtrace"] }
-bytes = "1.0.1"
 
 utils = { path = "../utils" }
-postgres_ffi = { path = "../postgres_ffi" }
 workspace_hack = { version = "0.1", path = "../../workspace_hack" }

libs/pageserver_api/src/lib.rs

@@ -2,7 +2,6 @@ use const_format::formatcp;
 
 /// Public API types
 pub mod models;
-pub mod reltag;
 
 pub const DEFAULT_PG_LISTEN_PORT: u16 = 64000;
 pub const DEFAULT_PG_LISTEN_ADDR: &str = formatcp!("127.0.0.1:{DEFAULT_PG_LISTEN_PORT}");

libs/pageserver_api/src/models.rs

@@ -7,10 +7,6 @@ use utils::{
     lsn::Lsn,
 };
 
-use crate::reltag::RelTag;
-use anyhow::bail;
-use bytes::{Buf, 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 {
@@ -23,22 +19,6 @@ pub enum TenantState {
     Broken,
 }
 
-/// 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.
-    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:
-    /// for example, if the owning tenant goes back to Active again.
-    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.
-    Broken,
-}
-
 #[serde_as]
 #[derive(Serialize, Deserialize)]
 pub struct TimelineCreateRequest {
@@ -143,15 +123,9 @@ pub struct TenantInfo {
     pub has_in_progress_downloads: Option<bool>,
 }
 
-/// This represents the output of the "timeline_detail" and "timeline_list" API calls.
 #[serde_as]
 #[derive(Debug, Serialize, Deserialize, Clone)]
-pub struct TimelineInfo {
-    #[serde_as(as = "DisplayFromStr")]
-    pub tenant_id: TenantId,
-    #[serde_as(as = "DisplayFromStr")]
-    pub timeline_id: TimelineId,
-
+pub struct LocalTimelineInfo {
     #[serde_as(as = "Option<DisplayFromStr>")]
     pub ancestor_timeline_id: Option<TimelineId>,
     #[serde_as(as = "Option<DisplayFromStr>")]
@@ -175,35 +149,28 @@ pub struct TimelineInfo {
     /// the timestamp (in microseconds) of the last received message
     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-
-    // compatility with older clients.
-    pub local: LocalTimelineInfo,
-    pub remote: RemoteTimelineInfo,
-}
-
-#[serde_as]
-#[derive(Debug, Serialize, Deserialize, Clone)]
-pub struct LocalTimelineInfo {
-    #[serde_as(as = "Option<DisplayFromStr>")]
-    pub ancestor_timeline_id: Option<TimelineId>,
-    #[serde_as(as = "Option<DisplayFromStr>")]
-    pub ancestor_lsn: Option<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
 }
 
 #[serde_as]
 #[derive(Debug, Serialize, Deserialize, Clone)]
 pub struct RemoteTimelineInfo {
-    #[serde_as(as = "Option<DisplayFromStr>")]
-    pub remote_consistent_lsn: Option<Lsn>,
+    #[serde_as(as = "DisplayFromStr")]
+    pub remote_consistent_lsn: Lsn,
+    pub awaits_download: bool,
+}
+
+///
+/// This represents the output of the "timeline_detail" API call.
+///
+#[serde_as]
+#[derive(Debug, Serialize, Deserialize, Clone)]
+pub struct TimelineInfo {
+    #[serde_as(as = "DisplayFromStr")]
+    pub tenant_id: TenantId,
+    #[serde_as(as = "DisplayFromStr")]
+    pub timeline_id: TimelineId,
+    pub local: Option<LocalTimelineInfo>,
+    pub remote: Option<RemoteTimelineInfo>,
 }
 
 pub type ConfigureFailpointsRequest = Vec<FailpointConfig>;
@@ -223,160 +190,3 @@ pub struct FailpointConfig {
 pub struct TimelineGcRequest {
     pub gc_horizon: Option<u64>,
 }
-
-// Wrapped in libpq CopyData
-pub enum PagestreamFeMessage {
-    Exists(PagestreamExistsRequest),
-    Nblocks(PagestreamNblocksRequest),
-    GetPage(PagestreamGetPageRequest),
-    DbSize(PagestreamDbSizeRequest),
-}
-
-// Wrapped in libpq CopyData
-pub enum PagestreamBeMessage {
-    Exists(PagestreamExistsResponse),
-    Nblocks(PagestreamNblocksResponse),
-    GetPage(PagestreamGetPageResponse),
-    Error(PagestreamErrorResponse),
-    DbSize(PagestreamDbSizeResponse),
-}
-
-#[derive(Debug)]
-pub struct PagestreamExistsRequest {
-    pub latest: bool,
-    pub lsn: Lsn,
-    pub rel: RelTag,
-}
-
-#[derive(Debug)]
-pub struct PagestreamNblocksRequest {
-    pub latest: bool,
-    pub lsn: Lsn,
-    pub rel: RelTag,
-}
-
-#[derive(Debug)]
-pub struct PagestreamGetPageRequest {
-    pub latest: bool,
-    pub lsn: Lsn,
-    pub rel: RelTag,
-    pub blkno: u32,
-}
-
-#[derive(Debug)]
-pub struct PagestreamDbSizeRequest {
-    pub latest: bool,
-    pub lsn: Lsn,
-    pub dbnode: u32,
-}
-
-#[derive(Debug)]
-pub struct PagestreamExistsResponse {
-    pub exists: bool,
-}
-
-#[derive(Debug)]
-pub struct PagestreamNblocksResponse {
-    pub n_blocks: u32,
-}
-
-#[derive(Debug)]
-pub struct PagestreamGetPageResponse {
-    pub page: Bytes,
-}
-
-#[derive(Debug)]
-pub struct PagestreamErrorResponse {
-    pub message: String,
-}
-
-#[derive(Debug)]
-pub struct PagestreamDbSizeResponse {
-    pub db_size: i64,
-}
-
-impl PagestreamFeMessage {
-    pub fn parse(mut body: Bytes) -> anyhow::Result<PagestreamFeMessage> {
-        // TODO these gets can fail
-
-        // these correspond to the NeonMessageTag enum in pagestore_client.h
-        //
-        // TODO: consider using protobuf or serde bincode for less error prone
-        // serialization.
-        let msg_tag = body.get_u8();
-        match msg_tag {
-            0 => Ok(PagestreamFeMessage::Exists(PagestreamExistsRequest {
-                latest: body.get_u8() != 0,
-                lsn: Lsn::from(body.get_u64()),
-                rel: RelTag {
-                    spcnode: body.get_u32(),
-                    dbnode: body.get_u32(),
-                    relnode: body.get_u32(),
-                    forknum: body.get_u8(),
-                },
-            })),
-            1 => Ok(PagestreamFeMessage::Nblocks(PagestreamNblocksRequest {
-                latest: body.get_u8() != 0,
-                lsn: Lsn::from(body.get_u64()),
-                rel: RelTag {
-                    spcnode: body.get_u32(),
-                    dbnode: body.get_u32(),
-                    relnode: body.get_u32(),
-                    forknum: body.get_u8(),
-                },
-            })),
-            2 => Ok(PagestreamFeMessage::GetPage(PagestreamGetPageRequest {
-                latest: body.get_u8() != 0,
-                lsn: Lsn::from(body.get_u64()),
-                rel: RelTag {
-                    spcnode: body.get_u32(),
-                    dbnode: body.get_u32(),
-                    relnode: body.get_u32(),
-                    forknum: body.get_u8(),
-                },
-                blkno: body.get_u32(),
-            })),
-            3 => Ok(PagestreamFeMessage::DbSize(PagestreamDbSizeRequest {
-                latest: body.get_u8() != 0,
-                lsn: Lsn::from(body.get_u64()),
-                dbnode: body.get_u32(),
-            })),
-            _ => bail!("unknown smgr message tag: {},'{:?}'", msg_tag, body),
-        }
-    }
-}
-
-impl PagestreamBeMessage {
-    pub fn serialize(&self) -> Bytes {
-        let mut bytes = BytesMut::new();
-
-        match self {
-            Self::Exists(resp) => {
-                bytes.put_u8(100); /* tag from pagestore_client.h */
-                bytes.put_u8(resp.exists as u8);
-            }
-
-            Self::Nblocks(resp) => {
-                bytes.put_u8(101); /* tag from pagestore_client.h */
-                bytes.put_u32(resp.n_blocks);
-            }
-
-            Self::GetPage(resp) => {
-                bytes.put_u8(102); /* tag from pagestore_client.h */
-                bytes.put(&resp.page[..]);
-            }
-
-            Self::Error(resp) => {
-                bytes.put_u8(103); /* tag from pagestore_client.h */
-                bytes.put(resp.message.as_bytes());
-                bytes.put_u8(0); // null terminator
-            }
-            Self::DbSize(resp) => {
-                bytes.put_u8(104); /* tag from pagestore_client.h */
-                bytes.put_i64(resp.db_size);
-            }
-        }
-
-        bytes.into()
-    }
-}

libs/postgres_ffi/Cargo.toml

@@ -13,7 +13,7 @@ crc32c = "0.6.0"
 hex = "0.4.3"
 once_cell = "1.13.0"
 log = "0.4.14"
-memoffset = "0.7"
+memoffset = "0.6.2"
 thiserror = "1.0"
 serde = { version = "1.0", features = ["derive"] }
 utils = { path = "../utils" }
@@ -26,4 +26,4 @@ wal_craft = { path = "wal_craft" }
 
 [build-dependencies]
 anyhow = "1.0"
-bindgen = "0.61"
+bindgen = "0.60.1"

libs/postgres_ffi/wal_craft/Cargo.toml

@@ -7,7 +7,7 @@ edition = "2021"
 
 [dependencies]
 anyhow = "1.0"
-clap = "4.0"
+clap = "3.0"
 env_logger = "0.9"
 log = "0.4"
 once_cell = "1.13.0"

libs/postgres_ffi/wal_craft/src/bin/wal_craft.rs

@@ -1,19 +1,68 @@
 use anyhow::*;
-use clap::{value_parser, Arg, ArgMatches, Command};
-use std::{path::PathBuf, str::FromStr};
+use clap::{App, Arg, ArgMatches};
+use std::str::FromStr;
 use wal_craft::*;
 
 fn main() -> Result<()> {
     env_logger::Builder::from_env(env_logger::Env::default().default_filter_or("wal_craft=info"))
         .init();
-    let arg_matches = cli().get_matches();
+    let type_arg = &Arg::new("type")
+        .takes_value(true)
+        .help("Type of WAL to craft")
+        .possible_values([
+            Simple::NAME,
+            LastWalRecordXlogSwitch::NAME,
+            LastWalRecordXlogSwitchEndsOnPageBoundary::NAME,
+            WalRecordCrossingSegmentFollowedBySmallOne::NAME,
+            LastWalRecordCrossingSegment::NAME,
+        ])
+        .required(true);
+    let arg_matches = App::new("Postgres WAL crafter")
+        .about("Crafts Postgres databases with specific WAL properties")
+        .subcommand(
+            App::new("print-postgres-config")
+                .about("Print the configuration required for PostgreSQL server before running this script")
+        )
+        .subcommand(
+            App::new("with-initdb")
+                .about("Craft WAL in a new data directory first initialized with initdb")
+                .arg(type_arg)
+                .arg(
+                    Arg::new("datadir")
+                        .takes_value(true)
+                        .help("Data directory for the Postgres server")
+                        .required(true)
+                )
+                .arg(
+                    Arg::new("pg-distrib-dir")
+                        .long("pg-distrib-dir")
+                        .takes_value(true)
+                        .help("Directory with Postgres distributions (bin and lib directories, e.g. pg_install containing subpath `v14/bin/postgresql`)")
+                        .default_value("/usr/local")
+                )
+                .arg(
+                    Arg::new("pg-version")
+                    .long("pg-version")
+                    .help("Postgres version to use for the initial tenant")
+                    .required(true)
+                    .takes_value(true)
+                )
+        )
+        .subcommand(
+            App::new("in-existing")
+                .about("Craft WAL at an existing recently created Postgres database. Note that server may append new WAL entries on shutdown.")
+                .arg(type_arg)
+                .arg(
+                    Arg::new("connection")
+                        .takes_value(true)
+                        .help("Connection string to the Postgres database to populate")
+                        .required(true)
+                )
+        )
+        .get_matches();
 
     let wal_craft = |arg_matches: &ArgMatches, client| {
-        let (intermediate_lsns, end_of_wal_lsn) = match arg_matches
-            .get_one::<String>("type")
-            .map(|s| s.as_str())
-            .context("'type' is required")?
-        {
+        let (intermediate_lsns, end_of_wal_lsn) = match arg_matches.value_of("type").unwrap() {
             Simple::NAME => Simple::craft(client)?,
             LastWalRecordXlogSwitch::NAME => LastWalRecordXlogSwitch::craft(client)?,
             LastWalRecordXlogSwitchEndsOnPageBoundary::NAME => {
@@ -23,12 +72,12 @@ fn main() -> Result<()> {
                 WalRecordCrossingSegmentFollowedBySmallOne::craft(client)?
             }
             LastWalRecordCrossingSegment::NAME => LastWalRecordCrossingSegment::craft(client)?,
-            a => panic!("Unknown --type argument: {a}"),
+            a => panic!("Unknown --type argument: {}", a),
         };
         for lsn in intermediate_lsns {
-            println!("intermediate_lsn = {lsn}");
+            println!("intermediate_lsn = {}", lsn);
         }
-        println!("end_of_wal = {end_of_wal_lsn}");
+        println!("end_of_wal = {}", end_of_wal_lsn);
         Ok(())
     };
 
@@ -36,24 +85,20 @@ fn main() -> Result<()> {
         None => panic!("No subcommand provided"),
         Some(("print-postgres-config", _)) => {
             for cfg in REQUIRED_POSTGRES_CONFIG.iter() {
-                println!("{cfg}");
+                println!("{}", cfg);
             }
             Ok(())
         }
 
         Some(("with-initdb", arg_matches)) => {
             let cfg = Conf {
-                pg_version: *arg_matches
-                    .get_one::<u32>("pg-version")
-                    .context("'pg-version' is required")?,
-                pg_distrib_dir: arg_matches
-                    .get_one::<PathBuf>("pg-distrib-dir")
-                    .context("'pg-distrib-dir' is required")?
-                    .to_owned(),
-                datadir: arg_matches
-                    .get_one::<PathBuf>("datadir")
-                    .context("'datadir' is required")?
-                    .to_owned(),
+                pg_version: arg_matches
+                    .value_of("pg-version")
+                    .unwrap()
+                    .parse::<u32>()
+                    .context("Failed to parse postgres version from the argument string")?,
+                pg_distrib_dir: arg_matches.value_of("pg-distrib-dir").unwrap().into(),
+                datadir: arg_matches.value_of("datadir").unwrap().into(),
             };
             cfg.initdb()?;
             let srv = cfg.start_server()?;
@@ -63,77 +108,9 @@ fn main() -> Result<()> {
         }
         Some(("in-existing", arg_matches)) => wal_craft(
             arg_matches,
-            &mut postgres::Config::from_str(
-                arg_matches
-                    .get_one::<String>("connection")
-                    .context("'connection' is required")?,
-            )
-            .context(
-                "'connection' argument value could not be parsed as a postgres connection string",
-            )?
-            .connect(postgres::NoTls)?,
+            &mut postgres::Config::from_str(arg_matches.value_of("connection").unwrap())?
+                .connect(postgres::NoTls)?,
         ),
         Some(_) => panic!("Unknown subcommand"),
     }
 }
-
-fn cli() -> Command {
-    let type_arg = &Arg::new("type")
-        .help("Type of WAL to craft")
-        .value_parser([
-            Simple::NAME,
-            LastWalRecordXlogSwitch::NAME,
-            LastWalRecordXlogSwitchEndsOnPageBoundary::NAME,
-            WalRecordCrossingSegmentFollowedBySmallOne::NAME,
-            LastWalRecordCrossingSegment::NAME,
-        ])
-        .required(true);
-
-    Command::new("Postgres WAL crafter")
-        .about("Crafts Postgres databases with specific WAL properties")
-        .subcommand(
-            Command::new("print-postgres-config")
-                .about("Print the configuration required for PostgreSQL server before running this script")
-        )
-        .subcommand(
-            Command::new("with-initdb")
-                .about("Craft WAL in a new data directory first initialized with initdb")
-                .arg(type_arg)
-                .arg(
-                    Arg::new("datadir")
-                        .help("Data directory for the Postgres server")
-                        .value_parser(value_parser!(PathBuf))
-                        .required(true)
-                )
-                .arg(
-                    Arg::new("pg-distrib-dir")
-                        .long("pg-distrib-dir")
-                        .value_parser(value_parser!(PathBuf))
-                        .help("Directory with Postgres distributions (bin and lib directories, e.g. pg_install containing subpath `v14/bin/postgresql`)")
-                        .default_value("/usr/local")
-                )
-                .arg(
-                    Arg::new("pg-version")
-                    .long("pg-version")
-                    .help("Postgres version to use for the initial tenant")
-                    .value_parser(value_parser!(u32))
-                    .required(true)
-
-                )
-        )
-        .subcommand(
-            Command::new("in-existing")
-                .about("Craft WAL at an existing recently created Postgres database. Note that server may append new WAL entries on shutdown.")
-                .arg(type_arg)
-                .arg(
-                    Arg::new("connection")
-                        .help("Connection string to the Postgres database to populate")
-                        .required(true)
-                )
-        )
-}
-
-#[test]
-fn verify_cli() {
-    cli().debug_assert();
-}

libs/postgres_ffi/wal_craft/src/lib.rs

@@ -37,22 +37,22 @@ pub static REQUIRED_POSTGRES_CONFIG: Lazy<Vec<&'static str>> = Lazy::new(|| {
 });
 
 impl Conf {
-    pub fn pg_distrib_dir(&self) -> anyhow::Result<PathBuf> {
+    pub fn pg_distrib_dir(&self) -> PathBuf {
         let path = self.pg_distrib_dir.clone();
 
         match self.pg_version {
-            14 => Ok(path.join(format!("v{}", self.pg_version))),
-            15 => Ok(path.join(format!("v{}", self.pg_version))),
-            _ => bail!("Unsupported postgres version: {}", self.pg_version),
+            14 => path.join(format!("v{}", self.pg_version)),
+            15 => path.join(format!("v{}", self.pg_version)),
+            _ => panic!("Unsupported postgres version: {}", self.pg_version),
         }
     }
 
-    fn pg_bin_dir(&self) -> anyhow::Result<PathBuf> {
-        Ok(self.pg_distrib_dir()?.join("bin"))
+    fn pg_bin_dir(&self) -> PathBuf {
+        self.pg_distrib_dir().join("bin")
     }
 
-    fn pg_lib_dir(&self) -> anyhow::Result<PathBuf> {
-        Ok(self.pg_distrib_dir()?.join("lib"))
+    fn pg_lib_dir(&self) -> PathBuf {
+        self.pg_distrib_dir().join("lib")
     }
 
     pub fn wal_dir(&self) -> PathBuf {
@@ -60,12 +60,12 @@ impl Conf {
     }
 
     fn new_pg_command(&self, command: impl AsRef<Path>) -> Result<Command> {
-        let path = self.pg_bin_dir()?.join(command);
+        let path = self.pg_bin_dir().join(command);
         ensure!(path.exists(), "Command {:?} does not exist", path);
         let mut cmd = Command::new(path);
         cmd.env_clear()
-            .env("LD_LIBRARY_PATH", self.pg_lib_dir()?)
-            .env("DYLD_LIBRARY_PATH", self.pg_lib_dir()?);
+            .env("LD_LIBRARY_PATH", self.pg_lib_dir())
+            .env("DYLD_LIBRARY_PATH", self.pg_lib_dir());
         Ok(cmd)
     }
 

libs/remote_storage/Cargo.toml

@@ -15,7 +15,7 @@ serde = { version = "1.0", features = ["derive"] }
 serde_json = "1"
 tokio = { version = "1.17", features = ["sync", "macros", "fs", "io-util"] }
 tokio-util = { version = "0.7", features = ["io"] }
-toml_edit = { version = "0.14", features = ["easy"] }
+toml_edit = { version = "0.13", features = ["easy"] }
 tracing = "0.1.27"
 
 workspace_hack = { version = "0.1", path = "../../workspace_hack" }

libs/remote_storage/src/local_fs.rs

@@ -16,7 +16,7 @@ use tokio::{
     io::{self, AsyncReadExt, AsyncSeekExt, AsyncWriteExt},
 };
 use tracing::*;
-use utils::crashsafe::path_with_suffix_extension;
+use utils::crashsafe_dir::path_with_suffix_extension;
 
 use crate::{Download, DownloadError, RemoteObjectId};
 

libs/safekeeper_api/Cargo.toml

@@ -5,7 +5,7 @@ edition = "2021"
 
 [dependencies]
 serde = { version = "1.0", features = ["derive"] }
-serde_with = "2.0"
+serde_with = "1.12.0"
 const_format = "0.2.21"
 
 utils = { path = "../utils" }

libs/safekeeper_api/src/models.rs

@@ -1,24 +1,8 @@
 use serde::{Deserialize, Serialize};
-use serde_with::{serde_as, DisplayFromStr};
+use utils::id::{NodeId, TimelineId};
 
-use utils::{
-    id::{NodeId, TenantId, TimelineId},
-    lsn::Lsn,
-};
-
-#[serde_as]
 #[derive(Serialize, Deserialize)]
 pub struct TimelineCreateRequest {
-    #[serde_as(as = "DisplayFromStr")]
-    pub tenant_id: TenantId,
-    #[serde_as(as = "DisplayFromStr")]
     pub timeline_id: TimelineId,
-    pub peer_ids: Option<Vec<NodeId>>,
-    pub pg_version: u32,
-    pub system_id: Option<u64>,
-    pub wal_seg_size: Option<u32>,
-    #[serde_as(as = "DisplayFromStr")]
-    pub commit_lsn: Lsn,
-    // If not passed, it is assigned to the beginning of commit_lsn segment.
-    pub local_start_lsn: Option<Lsn>,
+    pub peer_ids: Vec<NodeId>,
 }

libs/utils/Cargo.toml

@@ -19,8 +19,8 @@ thiserror = "1.0"
 tokio = { version = "1.17", features = ["macros"]}
 tokio-rustls = "0.23"
 tracing = "0.1"
-tracing-subscriber = { version = "0.3", features = ["env-filter", "json"] }
-nix = "0.25"
+tracing-subscriber = { version = "0.3", features = ["env-filter"] }
+nix = "0.23.0"
 signal-hook = "0.3.10"
 rand = "0.8.3"
 jsonwebtoken = "8"
@@ -28,10 +28,8 @@ hex = { version = "0.4.3", features = ["serde"] }
 rustls = "0.20.2"
 rustls-split = "0.3.0"
 git-version = "0.3.5"
-serde_with = "2.0"
+serde_with = "1.12.0"
 once_cell = "1.13.0"
-strum = "0.24"
-strum_macros = "0.24"
 
 
 metrics = { path = "../metrics" }
@@ -42,7 +40,7 @@ byteorder = "1.4.3"
 bytes = "1.0.1"
 hex-literal = "0.3"
 tempfile = "3.2"
-criterion = "0.4"
+criterion = "0.3"
 rustls-pemfile = "1"
 
 [[bench]]

libs/utils/src/crashsafe_dir.rs

@@ -12,8 +12,16 @@ pub fn create_dir(path: impl AsRef<Path>) -> io::Result<()> {
     let path = path.as_ref();
 
     fs::create_dir(path)?;
-    fsync_file_and_parent(path)?;
-    Ok(())
+    File::open(path)?.sync_all()?;
+
+    if let Some(parent) = path.parent() {
+        File::open(parent)?.sync_all()
+    } else {
+        Err(io::Error::new(
+            io::ErrorKind::InvalidInput,
+            "can't find parent",
+        ))
+    }
 }
 
 /// Similar to [`std::fs::create_dir_all`], except we fsync all
@@ -57,12 +65,12 @@ pub fn create_dir_all(path: impl AsRef<Path>) -> io::Result<()> {
 
     // Fsync the created directories from child to parent.
     for &path in dirs_to_create.iter() {
-        fsync(path)?;
+        File::open(path)?.sync_all()?;
     }
 
     // If we created any new directories, fsync the parent.
     if !dirs_to_create.is_empty() {
-        fsync(path)?;
+        File::open(path)?.sync_all()?;
     }
 
     Ok(())
@@ -84,33 +92,6 @@ pub fn path_with_suffix_extension(original_path: impl AsRef<Path>, suffix: &str)
         .with_extension(new_extension.as_ref())
 }
 
-pub fn fsync_file_and_parent(file_path: &Path) -> io::Result<()> {
-    let parent = file_path.parent().ok_or_else(|| {
-        io::Error::new(
-            io::ErrorKind::Other,
-            format!("File {file_path:?} has no parent"),
-        )
-    })?;
-
-    fsync(file_path)?;
-    fsync(parent)?;
-    Ok(())
-}
-
-pub fn fsync(path: &Path) -> io::Result<()> {
-    File::open(path)
-        .map_err(|e| io::Error::new(e.kind(), format!("Failed to open the file {path:?}: {e}")))
-        .and_then(|file| {
-            file.sync_all().map_err(|e| {
-                io::Error::new(
-                    e.kind(),
-                    format!("Failed to sync file {path:?} data and metadata: {e}"),
-                )
-            })
-        })
-        .map_err(|e| io::Error::new(e.kind(), format!("Failed to fsync file {path:?}: {e}")))
-}
-
 #[cfg(test)]
 mod tests {
     use tempfile::tempdir;

libs/utils/src/id.rs

@@ -75,12 +75,6 @@ impl From<[u8; 16]> for Id {
     }
 }
 
-impl From<Id> for u128 {
-    fn from(id: Id) -> Self {
-        u128::from_le_bytes(id.0)
-    }
-}
-
 impl fmt::Display for Id {
     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
         f.write_str(&self.hex_encode())
@@ -142,12 +136,6 @@ macro_rules! id_newtype {
             }
         }
 
-        impl From<$t> for u128 {
-            fn from(id: $t) -> Self {
-                u128::from(id.0)
-            }
-        }
-
         impl fmt::Display for $t {
             fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
                 self.0.fmt(f)

libs/utils/src/lib.rs

@@ -22,8 +22,8 @@ pub mod pq_proto;
 // dealing with connstring parsing and handy access to it's parts
 pub mod connstring;
 
-// helper functions for creating and fsyncing
-pub mod crashsafe;
+// helper functions for creating and fsyncing directories/trees
+pub mod crashsafe_dir;
 
 // common authentication routines
 pub mod auth;

libs/utils/src/logging.rs

@@ -1,35 +1,11 @@
 use std::{
     fs::{File, OpenOptions},
     path::Path,
-    str::FromStr,
 };
 
 use anyhow::{Context, Result};
-use strum_macros::{EnumString, EnumVariantNames};
 
-#[derive(EnumString, EnumVariantNames, Eq, PartialEq, Debug, Clone, Copy)]
-#[strum(serialize_all = "snake_case")]
-pub enum LogFormat {
-    Plain,
-    Json,
-}
-
-impl LogFormat {
-    pub fn from_config(s: &str) -> anyhow::Result<LogFormat> {
-        use strum::VariantNames;
-        LogFormat::from_str(s).with_context(|| {
-            format!(
-                "Unrecognized log format. Please specify one of: {:?}",
-                LogFormat::VARIANTS
-            )
-        })
-    }
-}
-pub fn init(
-    log_filename: impl AsRef<Path>,
-    daemonize: bool,
-    log_format: LogFormat,
-) -> Result<File> {
+pub fn init(log_filename: impl AsRef<Path>, daemonize: bool) -> Result<File> {
     // Don't open the same file for output multiple times;
     // the different fds could overwrite each other's output.
     let log_file = OpenOptions::new()
@@ -45,50 +21,22 @@ pub fn init(
     let env_filter = tracing_subscriber::EnvFilter::try_from_default_env()
         .unwrap_or_else(|_| tracing_subscriber::EnvFilter::new(default_filter_str));
 
-    let x: File = log_file.try_clone().unwrap();
     let base_logger = tracing_subscriber::fmt()
         .with_env_filter(env_filter)
-        .with_target(false)
-        .with_ansi(false)
-        .with_writer(move || -> Box<dyn std::io::Write> {
-            // we are cloning and returning log file in order to allow redirecting daemonized stdout and stderr to it
-            // if we do not use daemonization (e.g. in docker) it is better to log to stdout directly
-            // for example to be in line with docker log command which expects logs comimg from stdout
-            if daemonize {
-                Box::new(x.try_clone().unwrap())
-            } else {
-                Box::new(std::io::stdout())
-            }
-        });
+        .with_target(false) // don't include event targets
+        .with_ansi(false); // don't use colors in log file;
 
-    match log_format {
-        LogFormat::Json => base_logger.json().init(),
-        LogFormat::Plain => base_logger.init(),
+    // we are cloning and returning log file in order to allow redirecting daemonized stdout and stderr to it
+    // if we do not use daemonization (e.g. in docker) it is better to log to stdout directly
+    // for example to be in line with docker log command which expects logs comimg from stdout
+    if daemonize {
+        let x = log_file.try_clone().unwrap();
+        base_logger
+            .with_writer(move || x.try_clone().unwrap())
+            .init();
+    } else {
+        base_logger.init();
     }
 
     Ok(log_file)
 }
-
-// #[cfg(test)]
-// Due to global logger, can't run tests in same process.
-// So until there's a non-global one, the tests are in ../tests/ as separate files.
-#[macro_export(local_inner_macros)]
-macro_rules! test_init_file_logger {
-    ($log_level:expr, $log_format:expr) => {{
-        use std::str::FromStr;
-        std::env::set_var("RUST_LOG", $log_level);
-
-        let tmp_dir = tempfile::TempDir::new().unwrap();
-        let log_file_path = tmp_dir.path().join("logfile");
-
-        let log_format = $crate::logging::LogFormat::from_str($log_format).unwrap();
-        let _log_file = $crate::logging::init(&log_file_path, true, log_format).unwrap();
-
-        let log_file = std::fs::OpenOptions::new()
-            .read(true)
-            .open(&log_file_path)
-            .unwrap();
-
-        log_file
-    }};
-}

libs/utils/src/lsn.rs

@@ -66,11 +66,6 @@ impl Lsn {
         (self.0 % seg_sz as u64) as usize
     }
 
-    /// Compute LSN of the segment start.
-    pub fn segment_lsn(self, seg_sz: usize) -> Lsn {
-        Lsn(self.0 - (self.0 % seg_sz as u64))
-    }
-
     /// Compute the segment number
     pub fn segment_number(self, seg_sz: usize) -> u64 {
         self.0 / seg_sz as u64

libs/utils/src/postgres_backend_async.rs

@@ -15,7 +15,7 @@ use std::sync::Arc;
 use std::task::Poll;
 use tracing::{debug, error, trace};
 
-use tokio::io::{AsyncRead, AsyncWrite, AsyncWriteExt, BufReader};
+use tokio::io::{AsyncRead, AsyncWrite, AsyncWriteExt};
 use tokio_rustls::TlsAcceptor;
 
 #[async_trait::async_trait]
@@ -66,8 +66,8 @@ pub enum ProcessMsgResult {
 /// Always-writeable sock_split stream.
 /// May not be readable. See [`PostgresBackend::take_stream_in`]
 pub enum Stream {
-    Unencrypted(BufReader<tokio::net::TcpStream>),
-    Tls(Box<tokio_rustls::server::TlsStream<BufReader<tokio::net::TcpStream>>>),
+    Unencrypted(tokio::net::TcpStream),
+    Tls(Box<tokio_rustls::server::TlsStream<tokio::net::TcpStream>>),
     Broken,
 }
 
@@ -157,7 +157,7 @@ impl PostgresBackend {
         let peer_addr = socket.peer_addr()?;
 
         Ok(Self {
-            stream: Stream::Unencrypted(BufReader::new(socket)),
+            stream: Stream::Unencrypted(socket),
             buf_out: BytesMut::with_capacity(10 * 1024),
             state: ProtoState::Initialization,
             md5_salt: [0u8; 4],

libs/utils/tests/logger_json_test.rs

@@ -1,36 +0,0 @@
-// This could be in ../src/logging.rs but since the logger is global, these
-// can't be run in threads of the same process
-use std::fs::File;
-use std::io::{BufRead, BufReader, Lines};
-use tracing::*;
-use utils::test_init_file_logger;
-
-fn read_lines(file: File) -> Lines<BufReader<File>> {
-    BufReader::new(file).lines()
-}
-
-#[test]
-fn test_json_format_has_message_and_custom_field() {
-    std::env::set_var("RUST_LOG", "info");
-
-    let log_file = test_init_file_logger!("info", "json");
-
-    let custom_field: &str = "hi";
-    trace!(custom = %custom_field, "test log message");
-    debug!(custom = %custom_field, "test log message");
-    info!(custom = %custom_field, "test log message");
-    warn!(custom = %custom_field, "test log message");
-    error!(custom = %custom_field, "test log message");
-
-    let lines = read_lines(log_file);
-    for line in lines {
-        let content = line.unwrap();
-        let json_object = serde_json::from_str::<serde_json::Value>(&content).unwrap();
-
-        assert_eq!(json_object["fields"]["custom"], "hi");
-        assert_eq!(json_object["fields"]["message"], "test log message");
-
-        assert_ne!(json_object["level"], "TRACE");
-        assert_ne!(json_object["level"], "DEBUG");
-    }
-}

libs/utils/tests/logger_plain_test.rs

@@ -1,36 +0,0 @@
-// This could be in ../src/logging.rs but since the logger is global, these
-// can't be run in threads of the same process
-use std::fs::File;
-use std::io::{BufRead, BufReader, Lines};
-use tracing::*;
-use utils::test_init_file_logger;
-
-fn read_lines(file: File) -> Lines<BufReader<File>> {
-    BufReader::new(file).lines()
-}
-
-#[test]
-fn test_plain_format_has_message_and_custom_field() {
-    std::env::set_var("RUST_LOG", "warn");
-
-    let log_file = test_init_file_logger!("warn", "plain");
-
-    let custom_field: &str = "hi";
-    trace!(custom = %custom_field, "test log message");
-    debug!(custom = %custom_field, "test log message");
-    info!(custom = %custom_field, "test log message");
-    warn!(custom = %custom_field, "test log message");
-    error!(custom = %custom_field, "test log message");
-
-    let lines = read_lines(log_file);
-    for line in lines {
-        let content = line.unwrap();
-        serde_json::from_str::<serde_json::Value>(&content).unwrap_err();
-        assert!(content.contains("custom=hi"));
-        assert!(content.contains("test log message"));
-
-        assert!(!content.contains("TRACE"));
-        assert!(!content.contains("DEBUG"));
-        assert!(!content.contains("INFO"));
-    }
-}

pageserver/Cargo.toml

@@ -23,7 +23,7 @@ futures = "0.3.13"
 hex = "0.4.3"
 hyper = "0.14"
 itertools = "0.10.3"
-clap = { version = "4.0", features = ["string"] }
+clap = "3.0"
 daemonize = "0.4.1"
 tokio = { version = "1.17", features = ["process", "sync", "macros", "fs", "rt", "io-util", "time"] }
 tokio-util = { version = "0.7.3", features = ["io", "io-util"] }
@@ -38,25 +38,25 @@ tar = "0.4.33"
 humantime = "2.1.0"
 serde = { version = "1.0", features = ["derive"] }
 serde_json = "1"
-serde_with = "2.0"
+serde_with = "1.12.0"
 humantime-serde = "1.1.1"
 
 pprof = { git = "https://github.com/neondatabase/pprof-rs.git", branch = "wallclock-profiling", features = ["flamegraph"], optional = true }
 
-toml_edit = { version = "0.14", features = ["easy"] }
+toml_edit = { version = "0.13", features = ["easy"] }
 scopeguard = "1.1.0"
 const_format = "0.2.21"
 tracing = "0.1.36"
 signal-hook = "0.3.10"
 url = "2"
-nix = "0.25"
+nix = "0.23"
 once_cell = "1.13.0"
 crossbeam-utils = "0.8.5"
 fail = "0.5.0"
 git-version = "0.3.5"
 rstar = "0.9.3"
 num-traits = "0.2.15"
-amplify_num = { git = "https://github.com/hlinnaka/rust-amplify.git", branch = "unsigned-int-perf" }
+amplify_num = "0.4.1"
 
 pageserver_api = { path = "../libs/pageserver_api" }
 postgres_ffi = { path = "../libs/postgres_ffi" }
@@ -67,13 +67,7 @@ remote_storage = { path = "../libs/remote_storage" }
 workspace_hack = { version = "0.1", path = "../workspace_hack" }
 close_fds = "0.3.2"
 walkdir = "2.3.2"
-svg_fmt = "0.4.1"
 
 [dev-dependencies]
-criterion = "0.4"
 hex-literal = "0.3"
 tempfile = "3.2"
-
-[[bench]]
-name = "bench_layer_map"
-harness = false

pageserver/src/basebackup.rs

@@ -22,8 +22,8 @@ use std::time::SystemTime;
 use tar::{Builder, EntryType, Header};
 use tracing::*;
 
+use crate::reltag::{RelTag, SlruKind};
 use crate::tenant::Timeline;
-use pageserver_api::reltag::{RelTag, SlruKind};
 
 use postgres_ffi::pg_constants::{DEFAULTTABLESPACE_OID, GLOBALTABLESPACE_OID};
 use postgres_ffi::pg_constants::{PGDATA_SPECIAL_FILES, PGDATA_SUBDIRS, PG_HBA};

pageserver/src/bin/draw_timeline_dir.rs

@@ -1,150 +0,0 @@
-//! A tool for visualizing the arrangement of layerfiles within a timeline.
-//!
-//! It reads filenames from stdin and prints a svg on stdout. The image is a plot in
-//! page-lsn space, where every delta layer is a rectangle and every image layer is a
-//! thick line. Legend:
-//! - The x axis (left to right) represents page index.
-//! - The y axis represents LSN, growing upwards.
-//!
-//! Coordinates in both axis are compressed for better readability.
-//! (see https://medium.com/algorithms-digest/coordinate-compression-2fff95326fb)
-//!
-//! 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
-//! $ firefox out.svg
-//! ```
-//!
-//! This API was chosen so that we can easily work with filenames extracted from ssh,
-//! or from pageserver log files.
-//!
-//! TODO Consider shipping this as a grafana panel plugin:
-//!      https://grafana.com/tutorials/build-a-panel-plugin/
-use anyhow::Result;
-use pageserver::repository::Key;
-use std::cmp::Ordering;
-use std::io::{self, BufRead};
-use std::{
-    collections::{BTreeMap, BTreeSet},
-    ops::Range,
-};
-use svg_fmt::{rectangle, rgb, BeginSvg, EndSvg, Fill, Stroke};
-use utils::{lsn::Lsn, project_git_version};
-
-project_git_version!(GIT_VERSION);
-
-// Map values to their compressed coordinate - the index the value
-// would have in a sorted and deduplicated list of all values.
-fn build_coordinate_compression_map<T: Ord + Copy>(coords: Vec<T>) -> BTreeMap<T, usize> {
-    let set: BTreeSet<T> = coords.into_iter().collect();
-
-    let mut map: BTreeMap<T, usize> = BTreeMap::new();
-    for (i, e) in set.iter().enumerate() {
-        map.insert(*e, i);
-    }
-
-    map
-}
-
-fn parse_filename(name: &str) -> (Range<Key>, Range<Lsn>) {
-    let split: Vec<&str> = name.split("__").collect();
-    let keys: Vec<&str> = split[0].split('-').collect();
-    let mut lsns: Vec<&str> = split[1].split('-').collect();
-    if lsns.len() == 1 {
-        lsns.push(lsns[0]);
-    }
-
-    let keys = Key::from_hex(keys[0]).unwrap()..Key::from_hex(keys[1]).unwrap();
-    let lsns = Lsn::from_hex(lsns[0]).unwrap()..Lsn::from_hex(lsns[1]).unwrap();
-    (keys, lsns)
-}
-
-fn main() -> Result<()> {
-    // Parse layer filenames from stdin
-    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());
-        ranges.push(range);
-    }
-
-    // Collect all coordinates
-    let mut keys: Vec<Key> = vec![];
-    let mut lsns: Vec<Lsn> = vec![];
-    for (keyr, lsnr) in &ranges {
-        keys.push(keyr.start);
-        keys.push(keyr.end);
-        lsns.push(lsnr.start);
-        lsns.push(lsnr.end);
-    }
-
-    // Analyze
-    let key_map = build_coordinate_compression_map(keys);
-    let lsn_map = build_coordinate_compression_map(lsns);
-
-    // Initialize stats
-    let mut num_deltas = 0;
-    let mut num_images = 0;
-
-    // Draw
-    let stretch = 3.0; // Stretch out vertically for better visibility
-    println!(
-        "{}",
-        BeginSvg {
-            w: key_map.len() as f32,
-            h: stretch * lsn_map.len() as f32
-        }
-    );
-    for (keyr, lsnr) in &ranges {
-        let key_start = *key_map.get(&keyr.start).unwrap();
-        let key_end = *key_map.get(&keyr.end).unwrap();
-        let key_diff = key_end - key_start;
-        let lsn_max = lsn_map.len();
-
-        if key_start >= key_end {
-            panic!("Invalid key range {}-{}", key_start, key_end);
-        }
-
-        let lsn_start = *lsn_map.get(&lsnr.start).unwrap();
-        let lsn_end = *lsn_map.get(&lsnr.end).unwrap();
-
-        let mut lsn_diff = (lsn_end - lsn_start) as f32;
-        let mut fill = Fill::None;
-        let mut margin = 0.05 * lsn_diff; // Height-dependent margin to disambiguate overlapping deltas
-        let mut lsn_offset = 0.0;
-
-        // Fill in and thicken rectangle if it's an
-        // image layer so that we can see it.
-        match lsn_start.cmp(&lsn_end) {
-            Ordering::Less => num_deltas += 1,
-            Ordering::Equal => {
-                num_images += 1;
-                lsn_diff = 0.3;
-                lsn_offset = -lsn_diff / 2.0;
-                margin = 0.05;
-                fill = Fill::Color(rgb(0, 0, 0));
-            }
-            Ordering::Greater => panic!("Invalid lsn range {}-{}", lsn_start, lsn_end),
-        }
-
-        println!(
-            "    {}",
-            rectangle(
-                key_start as f32 + stretch * margin,
-                stretch * (lsn_max as f32 - (lsn_end as f32 - margin - lsn_offset)),
-                key_diff as f32 - stretch * 2.0 * margin,
-                stretch * (lsn_diff - 2.0 * margin)
-            )
-            .fill(fill)
-            .stroke(Stroke::Color(rgb(0, 0, 0), 0.1))
-            .border_radius(0.4)
-        );
-    }
-    println!("{}", EndSvg);
-
-    eprintln!("num_images: {}", num_images);
-    eprintln!("num_deltas: {}", num_deltas);
-
-    Ok(())
-}

pageserver/src/bin/pageserver.rs

@@ -6,7 +6,7 @@ use tracing::*;
 
 use anyhow::{anyhow, bail, Context, Result};
 
-use clap::{Arg, ArgAction, Command};
+use clap::{App, Arg};
 use daemonize::Daemonize;
 
 use fail::FailScenario;
@@ -14,7 +14,7 @@ use metrics::set_build_info_metric;
 
 use pageserver::{
     config::{defaults::*, PageServerConf},
-    http, page_cache, page_image_cache, page_service, profiling, task_mgr,
+    http, page_cache, page_service, profiling, task_mgr,
     task_mgr::TaskKind,
     task_mgr::{
         BACKGROUND_RUNTIME, COMPUTE_REQUEST_RUNTIME, MGMT_REQUEST_RUNTIME, WALRECEIVER_RUNTIME,
@@ -51,17 +51,57 @@ fn version() -> String {
 }
 
 fn main() -> anyhow::Result<()> {
-    let arg_matches = cli().get_matches();
+    let arg_matches = App::new("Neon page server")
+        .about("Materializes WAL stream to pages and serves them to the postgres")
+        .version(&*version())
+        .arg(
 
-    if arg_matches.get_flag("enabled-features") {
+            Arg::new("daemonize")
+                .short('d')
+                .long("daemonize")
+                .takes_value(false)
+                .help("Run in the background"),
+        )
+        .arg(
+            Arg::new("init")
+                .long("init")
+                .takes_value(false)
+                .help("Initialize pageserver with all given config overrides"),
+        )
+        .arg(
+            Arg::new("workdir")
+                .short('D')
+                .long("workdir")
+                .takes_value(true)
+                .help("Working directory for the pageserver"),
+        )
+        // See `settings.md` for more details on the extra configuration patameters pageserver can process
+        .arg(
+            Arg::new("config-override")
+                .short('c')
+                .takes_value(true)
+                .number_of_values(1)
+                .multiple_occurrences(true)
+                .help("Additional configuration overrides of the ones from the toml config file (or new ones to add there).
+                Any option has to be a valid toml document, example: `-c=\"foo='hey'\"` `-c=\"foo={value=1}\"`"),
+        )
+        .arg(Arg::new("update-config").long("update-config").takes_value(false).help(
+            "Update the config file when started",
+        ))
+        .arg(
+            Arg::new("enabled-features")
+                .long("enabled-features")
+                .takes_value(false)
+                .help("Show enabled compile time features"),
+        )
+        .get_matches();
+
+    if arg_matches.is_present("enabled-features") {
         println!("{{\"features\": {FEATURES:?} }}");
         return Ok(());
     }
 
-    let workdir = arg_matches
-        .get_one::<String>("workdir")
-        .map(Path::new)
-        .unwrap_or_else(|| Path::new(".neon"));
+    let workdir = Path::new(arg_matches.value_of("workdir").unwrap_or(".neon"));
     let workdir = workdir
         .canonicalize()
         .with_context(|| format!("Error opening workdir '{}'", workdir.display()))?;
@@ -75,7 +115,7 @@ fn main() -> anyhow::Result<()> {
         )
     })?;
 
-    let daemonize = arg_matches.get_flag("daemonize");
+    let daemonize = arg_matches.is_present("daemonize");
 
     let conf = match initialize_config(&cfg_file_path, arg_matches, &workdir)? {
         ControlFlow::Continue(conf) => conf,
@@ -87,7 +127,7 @@ fn main() -> anyhow::Result<()> {
 
     let tenants_path = conf.tenants_path();
     if !tenants_path.exists() {
-        utils::crashsafe::create_dir_all(conf.tenants_path()).with_context(|| {
+        utils::crashsafe_dir::create_dir_all(conf.tenants_path()).with_context(|| {
             format!(
                 "Failed to create tenants root dir at '{}'",
                 tenants_path.display()
@@ -101,7 +141,6 @@ fn main() -> anyhow::Result<()> {
     // Basic initialization of things that don't change after startup
     virtual_file::init(conf.max_file_descriptors);
     page_cache::init(conf.page_cache_size);
-    page_image_cache::init(64 * conf.page_cache_size); // temporary hack for benchmarking
 
     start_pageserver(conf, daemonize).context("Failed to start pageserver")?;
 
@@ -114,8 +153,8 @@ fn initialize_config(
     arg_matches: clap::ArgMatches,
     workdir: &Path,
 ) -> anyhow::Result<ControlFlow<(), &'static PageServerConf>> {
-    let init = arg_matches.get_flag("init");
-    let update_config = init || arg_matches.get_flag("update-config");
+    let init = arg_matches.is_present("init");
+    let update_config = init || arg_matches.is_present("update-config");
 
     let (mut toml, config_file_exists) = if cfg_file_path.is_file() {
         if init {
@@ -157,10 +196,13 @@ fn initialize_config(
         )
     };
 
-    if let Some(values) = arg_matches.get_many::<String>("config-override") {
+    if let Some(values) = arg_matches.values_of("config-override") {
         for option_line in values {
             let doc = toml_edit::Document::from_str(option_line).with_context(|| {
-                format!("Option '{option_line}' could not be parsed as a toml document")
+                format!(
+                    "Option '{}' could not be parsed as a toml document",
+                    option_line
+                )
             })?;
 
             for (key, item) in doc.iter() {
@@ -200,9 +242,9 @@ fn initialize_config(
 
 fn start_pageserver(conf: &'static PageServerConf, daemonize: bool) -> Result<()> {
     // Initialize logger
-    let log_file = logging::init(LOG_FILE_NAME, daemonize, conf.log_format)?;
+    let log_file = logging::init(LOG_FILE_NAME, daemonize)?;
 
-    info!("version: {}", version());
+    info!("version: {GIT_VERSION}");
 
     // TODO: Check that it looks like a valid repository before going further
 
@@ -343,55 +385,3 @@ fn start_pageserver(conf: &'static PageServerConf, daemonize: bool) -> Result<()
         }
     })
 }
-
-fn cli() -> Command {
-    Command::new("Neon page server")
-        .about("Materializes WAL stream to pages and serves them to the postgres")
-        .version(version())
-        .arg(
-
-            Arg::new("daemonize")
-                .short('d')
-                .long("daemonize")
-                .action(ArgAction::SetTrue)
-                .help("Run in the background"),
-        )
-        .arg(
-            Arg::new("init")
-                .long("init")
-                .action(ArgAction::SetTrue)
-                .help("Initialize pageserver with all given config overrides"),
-        )
-        .arg(
-            Arg::new("workdir")
-                .short('D')
-                .long("workdir")
-                .help("Working directory for the pageserver"),
-        )
-        // See `settings.md` for more details on the extra configuration patameters pageserver can process
-        .arg(
-            Arg::new("config-override")
-                .short('c')
-                .num_args(1)
-                .action(ArgAction::Append)
-                .help("Additional configuration overrides of the ones from the toml config file (or new ones to add there). \
-                Any option has to be a valid toml document, example: `-c=\"foo='hey'\"` `-c=\"foo={value=1}\"`"),
-        )
-        .arg(
-            Arg::new("update-config")
-                .long("update-config")
-                .action(ArgAction::SetTrue)
-                .help("Update the config file when started"),
-        )
-        .arg(
-            Arg::new("enabled-features")
-                .long("enabled-features")
-                .action(ArgAction::SetTrue)
-                .help("Show enabled compile time features"),
-        )
-}
-
-#[test]
-fn verify_cli() {
-    cli().debug_assert();
-}

pageserver/src/bin/pageserver_binutils.rs

@@ -9,7 +9,7 @@ use std::{
 };
 
 use anyhow::Context;
-use clap::{value_parser, Arg, Command};
+use clap::{App, Arg};
 
 use pageserver::{
     page_cache,
@@ -24,14 +24,40 @@ project_git_version!(GIT_VERSION);
 const METADATA_SUBCOMMAND: &str = "metadata";
 
 fn main() -> anyhow::Result<()> {
-    let arg_matches = cli().get_matches();
+    let arg_matches = App::new("Neon Pageserver binutils")
+        .about("Reads pageserver (and related) binary files management utility")
+        .version(GIT_VERSION)
+        .arg(Arg::new("path").help("Input file path").required(false))
+        .subcommand(
+            App::new(METADATA_SUBCOMMAND)
+                .about("Read and update pageserver metadata file")
+                .arg(
+                    Arg::new("metadata_path")
+                        .help("Input metadata file path")
+                        .required(false),
+                )
+                .arg(
+                    Arg::new("disk_consistent_lsn")
+                        .long("disk_consistent_lsn")
+                        .takes_value(true)
+                        .help("Replace disk consistent Lsn"),
+                )
+                .arg(
+                    Arg::new("prev_record_lsn")
+                        .long("prev_record_lsn")
+                        .takes_value(true)
+                        .help("Replace previous record Lsn"),
+                ),
+        )
+        .get_matches();
 
     match arg_matches.subcommand() {
         Some((subcommand_name, subcommand_matches)) => {
-            let path = subcommand_matches
-                .get_one::<PathBuf>("metadata_path")
-                .context("'metadata_path' argument is missing")?
-                .to_path_buf();
+            let path = PathBuf::from(
+                subcommand_matches
+                    .value_of("metadata_path")
+                    .context("'metadata_path' argument is missing")?,
+            );
             anyhow::ensure!(
                 subcommand_name == METADATA_SUBCOMMAND,
                 "Unknown subcommand {subcommand_name}"
@@ -39,10 +65,11 @@ fn main() -> anyhow::Result<()> {
             handle_metadata(&path, subcommand_matches)?;
         }
         None => {
-            let path = arg_matches
-                .get_one::<PathBuf>("path")
-                .context("'path' argument is missing")?
-                .to_path_buf();
+            let path = PathBuf::from(
+                arg_matches
+                    .value_of("path")
+                    .context("'path' argument is missing")?,
+            );
             println!(
                 "No subcommand specified, attempting to guess the format for file {}",
                 path.display()
@@ -83,7 +110,7 @@ fn handle_metadata(path: &Path, arg_matches: &clap::ArgMatches) -> Result<(), an
     let mut meta = TimelineMetadata::from_bytes(&metadata_bytes)?;
     println!("Current metadata:\n{meta:?}");
     let mut update_meta = false;
-    if let Some(disk_consistent_lsn) = arg_matches.get_one::<String>("disk_consistent_lsn") {
+    if let Some(disk_consistent_lsn) = arg_matches.value_of("disk_consistent_lsn") {
         meta = TimelineMetadata::new(
             Lsn::from_str(disk_consistent_lsn)?,
             meta.prev_record_lsn(),
@@ -95,7 +122,7 @@ fn handle_metadata(path: &Path, arg_matches: &clap::ArgMatches) -> Result<(), an
         );
         update_meta = true;
     }
-    if let Some(prev_record_lsn) = arg_matches.get_one::<String>("prev_record_lsn") {
+    if let Some(prev_record_lsn) = arg_matches.value_of("prev_record_lsn") {
         meta = TimelineMetadata::new(
             meta.disk_consistent_lsn(),
             Some(Lsn::from_str(prev_record_lsn)?),
@@ -115,40 +142,3 @@ fn handle_metadata(path: &Path, arg_matches: &clap::ArgMatches) -> Result<(), an
 
     Ok(())
 }
-
-fn cli() -> Command {
-    Command::new("Neon Pageserver binutils")
-        .about("Reads pageserver (and related) binary files management utility")
-        .version(GIT_VERSION)
-        .arg(
-            Arg::new("path")
-                .help("Input file path")
-                .value_parser(value_parser!(PathBuf))
-                .required(false),
-        )
-        .subcommand(
-            Command::new(METADATA_SUBCOMMAND)
-                .about("Read and update pageserver metadata file")
-                .arg(
-                    Arg::new("metadata_path")
-                        .help("Input metadata file path")
-                        .value_parser(value_parser!(PathBuf))
-                        .required(false),
-                )
-                .arg(
-                    Arg::new("disk_consistent_lsn")
-                        .long("disk_consistent_lsn")
-                        .help("Replace disk consistent Lsn"),
-                )
-                .arg(
-                    Arg::new("prev_record_lsn")
-                        .long("prev_record_lsn")
-                        .help("Replace previous record Lsn"),
-                ),
-        )
-}
-
-#[test]
-fn verify_cli() {
-    cli().debug_assert();
-}

pageserver/src/config.rs

@@ -7,7 +7,6 @@
 use anyhow::{anyhow, bail, ensure, Context, Result};
 use remote_storage::RemoteStorageConfig;
 use std::env;
-use utils::crashsafe::path_with_suffix_extension;
 
 use std::path::{Path, PathBuf};
 use std::str::FromStr;
@@ -17,7 +16,6 @@ use toml_edit::{Document, Item};
 use url::Url;
 use utils::{
     id::{NodeId, TenantId, TimelineId},
-    logging::LogFormat,
     postgres_backend::AuthType,
 };
 
@@ -26,7 +24,6 @@ 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";
 
 pub mod defaults {
@@ -46,8 +43,6 @@ pub mod defaults {
     pub const DEFAULT_PAGE_CACHE_SIZE: usize = 8192;
     pub const DEFAULT_MAX_FILE_DESCRIPTORS: usize = 100;
 
-    pub const DEFAULT_LOG_FORMAT: &str = "plain";
-
     ///
     /// Default built-in configuration file.
     ///
@@ -66,7 +61,6 @@ pub mod defaults {
 # initial superuser role name to use when creating a new tenant
 #initial_superuser_name = '{DEFAULT_SUPERUSER}'
 
-#log_format = '{DEFAULT_LOG_FORMAT}'
 # [tenant_config]
 #checkpoint_distance = {DEFAULT_CHECKPOINT_DISTANCE} # in bytes
 #checkpoint_timeout = {DEFAULT_CHECKPOINT_TIMEOUT}
@@ -130,8 +124,6 @@ pub struct PageServerConf {
 
     /// Etcd broker endpoints to connect to.
     pub broker_endpoints: Vec<Url>,
-
-    pub log_format: LogFormat,
 }
 
 #[derive(Debug, Clone, PartialEq, Eq)]
@@ -198,8 +190,6 @@ struct PageServerConfigBuilder {
     profiling: BuilderValue<ProfilingConfig>,
     broker_etcd_prefix: BuilderValue<String>,
     broker_endpoints: BuilderValue<Vec<Url>>,
-
-    log_format: BuilderValue<LogFormat>,
 }
 
 impl Default for PageServerConfigBuilder {
@@ -227,7 +217,6 @@ impl Default for PageServerConfigBuilder {
             profiling: Set(ProfilingConfig::Disabled),
             broker_etcd_prefix: Set(etcd_broker::DEFAULT_NEON_BROKER_ETCD_PREFIX.to_string()),
             broker_endpoints: Set(Vec::new()),
-            log_format: Set(LogFormat::from_str(DEFAULT_LOG_FORMAT).unwrap()),
         }
     }
 }
@@ -300,10 +289,6 @@ impl PageServerConfigBuilder {
         self.profiling = BuilderValue::Set(profiling)
     }
 
-    pub fn log_format(&mut self, log_format: LogFormat) {
-        self.log_format = BuilderValue::Set(log_format)
-    }
-
     pub fn build(self) -> anyhow::Result<PageServerConf> {
         let broker_endpoints = self
             .broker_endpoints
@@ -348,7 +333,6 @@ impl PageServerConfigBuilder {
             broker_etcd_prefix: self
                 .broker_etcd_prefix
                 .ok_or(anyhow!("missing broker_etcd_prefix"))?,
-            log_format: self.log_format.ok_or(anyhow!("missing log_format"))?,
         })
     }
 }
@@ -380,17 +364,6 @@ impl PageServerConf {
         self.timelines_path(tenant_id).join(timeline_id.to_string())
     }
 
-    pub fn timeline_uninit_mark_file_path(
-        &self,
-        tenant_id: TenantId,
-        timeline_id: TimelineId,
-    ) -> PathBuf {
-        path_with_suffix_extension(
-            self.timeline_path(&timeline_id, &tenant_id),
-            TIMELINE_UNINIT_MARK_SUFFIX,
-        )
-    }
-
     /// 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 {
@@ -401,28 +374,28 @@ impl PageServerConf {
     //
     // Postgres distribution paths
     //
-    pub fn pg_distrib_dir(&self, pg_version: u32) -> anyhow::Result<PathBuf> {
+    pub fn pg_distrib_dir(&self, pg_version: u32) -> PathBuf {
         let path = self.pg_distrib_dir.clone();
 
         match pg_version {
-            14 => Ok(path.join(format!("v{pg_version}"))),
-            15 => Ok(path.join(format!("v{pg_version}"))),
-            _ => bail!("Unsupported postgres version: {}", pg_version),
+            14 => path.join(format!("v{pg_version}")),
+            15 => path.join(format!("v{pg_version}")),
+            _ => panic!("Unsupported postgres version: {}", pg_version),
         }
     }
 
-    pub fn pg_bin_dir(&self, pg_version: u32) -> anyhow::Result<PathBuf> {
+    pub fn pg_bin_dir(&self, pg_version: u32) -> PathBuf {
         match pg_version {
-            14 => Ok(self.pg_distrib_dir(pg_version)?.join("bin")),
-            15 => Ok(self.pg_distrib_dir(pg_version)?.join("bin")),
-            _ => bail!("Unsupported postgres version: {}", pg_version),
+            14 => self.pg_distrib_dir(pg_version).join("bin"),
+            15 => self.pg_distrib_dir(pg_version).join("bin"),
+            _ => panic!("Unsupported postgres version: {}", pg_version),
         }
     }
-    pub fn pg_lib_dir(&self, pg_version: u32) -> anyhow::Result<PathBuf> {
+    pub fn pg_lib_dir(&self, pg_version: u32) -> PathBuf {
         match pg_version {
-            14 => Ok(self.pg_distrib_dir(pg_version)?.join("lib")),
-            15 => Ok(self.pg_distrib_dir(pg_version)?.join("lib")),
-            _ => bail!("Unsupported postgres version: {}", pg_version),
+            14 => self.pg_distrib_dir(pg_version).join("lib"),
+            15 => self.pg_distrib_dir(pg_version).join("lib"),
+            _ => panic!("Unsupported postgres version: {}", pg_version),
         }
     }
 
@@ -473,9 +446,6 @@ impl PageServerConf {
                         })
                         .collect::<anyhow::Result<_>>()?,
                 ),
-                "log_format" => builder.log_format(
-                    LogFormat::from_config(&parse_toml_string(key, item)?)?
-                ),
                 _ => bail!("unrecognized pageserver option '{key}'"),
             }
         }
@@ -588,7 +558,6 @@ impl PageServerConf {
             default_tenant_conf: TenantConf::dummy_conf(),
             broker_endpoints: Vec::new(),
             broker_etcd_prefix: etcd_broker::DEFAULT_NEON_BROKER_ETCD_PREFIX.to_string(),
-            log_format: LogFormat::from_str(defaults::DEFAULT_LOG_FORMAT).unwrap(),
         }
     }
 }
@@ -683,8 +652,6 @@ max_file_descriptors = 333
 initial_superuser_name = 'zzzz'
 id = 10
 
-log_format = 'json'
-
 "#;
 
     #[test]
@@ -724,7 +691,6 @@ log_format = 'json'
                     .parse()
                     .expect("Failed to parse a valid broker endpoint URL")],
                 broker_etcd_prefix: etcd_broker::DEFAULT_NEON_BROKER_ETCD_PREFIX.to_string(),
-                log_format: LogFormat::from_str(defaults::DEFAULT_LOG_FORMAT).unwrap(),
             },
             "Correct defaults should be used when no config values are provided"
         );
@@ -769,7 +735,6 @@ log_format = 'json'
                     .parse()
                     .expect("Failed to parse a valid broker endpoint URL")],
                 broker_etcd_prefix: etcd_broker::DEFAULT_NEON_BROKER_ETCD_PREFIX.to_string(),
-                log_format: LogFormat::Json,
             },
             "Should be able to parse all basic config values correctly"
         );

pageserver/src/http/openapi_spec.yml

@@ -1,11 +1,7 @@
 openapi: "3.0.2"
 info:
   title: Page Server API
-  description: Neon Pageserver API
   version: "1.0"
-  license:
-    name: "Apache"
-    url: https://github.com/neondatabase/neon/blob/main/LICENSE
 servers:
   - url: ""
 paths:
@@ -211,6 +207,7 @@ paths:
               schema:
                 $ref: "#/components/schemas/Error"
 
+ 
   /v1/tenant/{tenant_id}/timeline/{timeline_id}/get_lsn_by_timestamp:
     parameters:
       - name: tenant_id
@@ -615,10 +612,6 @@ components:
       required:
         - timeline_id
         - tenant_id
-        - last_record_lsn
-        - disk_consistent_lsn
-        - awaits_download
-        - state
       properties:
         timeline_id:
           type: string
@@ -626,15 +619,33 @@ components:
         tenant_id:
           type: string
           format: hex
+        local:
+          $ref: "#/components/schemas/LocalTimelineInfo"
+        remote:
+          $ref: "#/components/schemas/RemoteTimelineInfo"
+    RemoteTimelineInfo:
+      type: object
+      required:
+        - awaits_download
+        - remote_consistent_lsn
+      properties:
+        awaits_download:
+          type: boolean
+        remote_consistent_lsn:
+          type: string
+          format: hex
+    LocalTimelineInfo:
+      type: object
+      required:
+        - last_record_lsn
+        - disk_consistent_lsn
+      properties:
         last_record_lsn:
           type: string
           format: hex
         disk_consistent_lsn:
           type: string
           format: hex
-        remote_consistent_lsn:
-          type: string
-          format: hex
         ancestor_timeline_id:
           type: string
           format: hex
@@ -659,41 +670,7 @@ components:
           format: hex
         last_received_msg_ts:
           type: integer
-        awaits_download:
-          type: boolean
-        state:
-          type: string
 
-        # These 'local' and 'remote' fields just duplicate some of the fields
-        # above. They are kept for backwards-compatibility. They can be removed,
-        # when the control plane has been updated to look at the above fields
-        # directly.
-        local:
-          $ref: "#/components/schemas/LocalTimelineInfo"
-        remote:
-          $ref: "#/components/schemas/RemoteTimelineInfo"
-
-    LocalTimelineInfo:
-      type: object
-      properties:
-        ancestor_timeline_id:
-          type: string
-          format: hex
-        ancestor_lsn:
-          type: string
-          format: hex
-        current_logical_size:
-          type: integer
-        current_physical_size:
-          type: integer
-    RemoteTimelineInfo:
-      type: object
-      required:
-        - remote_consistent_lsn
-      properties:
-        remote_consistent_lsn:
-          type: string
-          format: hex
     Error:
       type: object
       required:

pageserver/src/http/routes.rs

@@ -79,13 +79,13 @@ fn get_config(request: &Request<Body>) -> &'static PageServerConf {
     get_state(request).conf
 }
 
-// Helper function to construct a TimelineInfo struct for a timeline
-async fn build_timeline_info(
-    state: &State,
+// Helper functions to construct a LocalTimelineInfo struct for a timeline
+
+fn local_timeline_info_from_timeline(
     timeline: &Arc<Timeline>,
     include_non_incremental_logical_size: bool,
     include_non_incremental_physical_size: bool,
-) -> anyhow::Result<TimelineInfo> {
+) -> anyhow::Result<LocalTimelineInfo> {
     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();
@@ -100,48 +100,24 @@ 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,
-        lsn @ Lsn(_) => Some(lsn),
-    };
-    let current_logical_size = match timeline.get_current_logical_size() {
-        Ok(size) => Some(size),
-        Err(err) => {
-            error!("Timeline info creation failed to get current logical size: {err:?}");
-            None
-        }
-    };
-    let current_physical_size = Some(timeline.get_physical_size());
-    let state = timeline.current_state();
-
-    let info = TimelineInfo {
-        tenant_id: timeline.tenant_id,
-        timeline_id: timeline.timeline_id,
-        ancestor_timeline_id,
-        ancestor_lsn,
+    let info = LocalTimelineInfo {
+        ancestor_timeline_id: timeline.get_ancestor_timeline_id(),
+        ancestor_lsn: {
+            match timeline.get_ancestor_lsn() {
+                Lsn(0) => None,
+                lsn @ Lsn(_) => Some(lsn),
+            }
+        },
         disk_consistent_lsn: timeline.get_disk_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: Some(
+            timeline
+                .get_current_logical_size()
+                .context("Timeline info creation failed to get current logical size")?,
+        ),
+        current_physical_size: Some(timeline.get_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 {
@@ -156,26 +132,32 @@ async fn build_timeline_info(
         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
-        // with the control plane.
-        local: LocalTimelineInfo {
-            ancestor_timeline_id,
-            ancestor_lsn,
-            current_logical_size,
-            current_physical_size,
-        },
-        remote: RemoteTimelineInfo {
-            remote_consistent_lsn,
-        },
     };
     Ok(info)
 }
 
+fn list_local_timelines(
+    tenant_id: TenantId,
+    include_non_incremental_logical_size: bool,
+    include_non_incremental_physical_size: bool,
+) -> Result<Vec<(TimelineId, LocalTimelineInfo)>> {
+    let tenant = tenant_mgr::get_tenant(tenant_id, true)?;
+    let timelines = tenant.list_timelines();
+
+    let mut local_timeline_info = Vec::with_capacity(timelines.len());
+    for (timeline_id, repository_timeline) in timelines {
+        local_timeline_info.push((
+            timeline_id,
+            local_timeline_info_from_timeline(
+                &repository_timeline,
+                include_non_incremental_logical_size,
+                include_non_incremental_physical_size,
+            )?,
+        ))
+    }
+    Ok(local_timeline_info)
+}
+
 // healthcheck handler
 async fn status_handler(request: Request<Body>) -> Result<Response<Body>, ApiError> {
     let config = get_config(&request);
@@ -187,8 +169,6 @@ 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 tenant = tenant_mgr::get_tenant(tenant_id, true).map_err(ApiError::NotFound)?;
     let new_timeline_info = async {
         match tenant.create_timeline(
@@ -199,10 +179,14 @@ async fn timeline_create_handler(mut request: Request<Body>) -> Result<Response<
         ).await {
             Ok(Some(new_timeline)) => {
                 // Created. Construct a TimelineInfo for it.
-                let timeline_info = build_timeline_info(state, &new_timeline, false, false)
-                    .await
+                let local_info = local_timeline_info_from_timeline(&new_timeline, false, false)
                     .map_err(ApiError::InternalServerError)?;
-                Ok(Some(timeline_info))
+                Ok(Some(TimelineInfo {
+                    tenant_id,
+                    timeline_id: new_timeline.timeline_id,
+                    local: Some(local_info),
+                    remote: None,
+                }))
             }
             Ok(None) => Ok(None), // timeline already exists
             Err(err) => Err(ApiError::InternalServerError(err)),
@@ -225,8 +209,6 @@ 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 timelines = tokio::task::spawn_blocking(move || {
         let _enter = info_span!("timeline_list", tenant = %tenant_id).entered();
         let tenant = tenant_mgr::get_tenant(tenant_id, true).map_err(ApiError::NotFound)?;
@@ -236,18 +218,36 @@ async fn timeline_list_handler(request: Request<Body>) -> Result<Response<Body>,
     .map_err(|e: JoinError| ApiError::InternalServerError(e.into()))??;
 
     let mut response_data = Vec::with_capacity(timelines.len());
-    for timeline in timelines {
-        let timeline_info = build_timeline_info(
-            state,
+    for (timeline_id, timeline) in timelines {
+        let local = match local_timeline_info_from_timeline(
             &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)?;
+        ) {
+            Ok(local) => Some(local),
+            Err(e) => {
+                error!("Failed to convert tenant timeline {timeline_id} into the local one: {e:?}");
+                None
+            }
+        };
 
-        response_data.push(timeline_info);
+        response_data.push(TimelineInfo {
+            tenant_id,
+            timeline_id,
+            local,
+            remote: get_state(&request)
+                .remote_index
+                .read()
+                .await
+                .timeline_entry(&TenantTimelineId {
+                    tenant_id,
+                    timeline_id,
+                })
+                .map(|remote_entry| RemoteTimelineInfo {
+                    remote_consistent_lsn: remote_entry.metadata.disk_consistent_lsn(),
+                    awaits_download: remote_entry.awaits_download,
+                }),
+        })
     }
 
     json_response(StatusCode::OK, response_data)
@@ -292,33 +292,59 @@ 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 (local_timeline_info, remote_timeline_info) = async {
         let timeline = tokio::task::spawn_blocking(move || {
-            tenant_mgr::get_tenant(tenant_id, true)?.get_timeline(timeline_id, false)
+            tenant_mgr::get_tenant(tenant_id, true)?.get_timeline(timeline_id)
         })
         .await
         .map_err(|e: JoinError| ApiError::InternalServerError(e.into()))?;
 
-        let timeline = timeline.map_err(ApiError::NotFound)?;
+        let local_timeline_info = match timeline.and_then(|timeline| {
+            local_timeline_info_from_timeline(
+                &timeline,
+                include_non_incremental_logical_size,
+                include_non_incremental_physical_size,
+            )
+        }) {
+            Ok(local_info) => Some(local_info),
+            Err(e) => {
+                error!("Failed to get local timeline info: {e:#}");
+                None
+            }
+        };
 
-        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)?;
-
-        Ok::<_, ApiError>(timeline_info)
+        let remote_timeline_info = {
+            let remote_index_read = get_state(&request).remote_index.read().await;
+            remote_index_read
+                .timeline_entry(&TenantTimelineId {
+                    tenant_id,
+                    timeline_id,
+                })
+                .map(|remote_entry| RemoteTimelineInfo {
+                    remote_consistent_lsn: remote_entry.metadata.disk_consistent_lsn(),
+                    awaits_download: remote_entry.awaits_download,
+                })
+        };
+        Ok::<_, ApiError>((local_timeline_info, remote_timeline_info))
     }
     .instrument(info_span!("timeline_detail", tenant = %tenant_id, timeline = %timeline_id))
     .await?;
 
-    json_response(StatusCode::OK, timeline_info)
+    if local_timeline_info.is_none() && remote_timeline_info.is_none() {
+        Err(ApiError::NotFound(anyhow!(
+            "Timeline {tenant_id}/{timeline_id} is not found neither locally nor remotely"
+        )))
+    } else {
+        json_response(
+            StatusCode::OK,
+            TimelineInfo {
+                tenant_id,
+                timeline_id,
+                local: local_timeline_info,
+                remote: remote_timeline_info,
+            },
+        )
+    }
 }
 
 async fn get_lsn_by_timestamp_handler(request: Request<Body>) -> Result<Response<Body>, ApiError> {
@@ -333,13 +359,14 @@ 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)
-        .and_then(|tenant| tenant.get_timeline(timeline_id, true))
+        .and_then(|tenant| tenant.get_timeline(timeline_id))
+        .with_context(|| format!("No timeline {timeline_id} in repository for tenant {tenant_id}"))
         .map_err(ApiError::NotFound)?;
     let result = match timeline
         .find_lsn_for_timestamp(timestamp_pg)
         .map_err(ApiError::InternalServerError)?
     {
-        LsnForTimestamp::Present(lsn) => format!("{lsn}"),
+        LsnForTimestamp::Present(lsn) => format!("{}", lsn),
         LsnForTimestamp::Future(_lsn) => "future".into(),
         LsnForTimestamp::Past(_lsn) => "past".into(),
         LsnForTimestamp::NoData(_lsn) => "nodata".into(),
@@ -387,7 +414,7 @@ async fn tenant_attach_handler(request: Request<Body>) -> Result<Response<Body>,
         }
         return json_response(StatusCode::ACCEPTED, ());
     }
-    // no tenant in the index, release the lock to make the potentially lengthy download operation
+    // no tenant in the index, release the lock to make the potentially lengthy download opetation
     drop(index_accessor);
 
     // download index parts for every tenant timeline
@@ -539,27 +566,36 @@ async fn tenant_status(request: Request<Body>) -> Result<Response<Body>, ApiErro
             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))
-        }
+    let tenant_state = match tenant {
+        Ok(tenant) => tenant.current_state(),
         Err(e) => {
             error!("Failed to get local tenant state: {e:#}");
             if has_in_progress_downloads {
-                (TenantState::Paused, None)
+                TenantState::Paused
             } else {
-                (TenantState::Broken, None)
+                TenantState::Broken
             }
         }
     };
 
+    let current_physical_size =
+        match tokio::task::spawn_blocking(move || list_local_timelines(tenant_id, false, false))
+            .await
+            .map_err(|e: JoinError| ApiError::InternalServerError(e.into()))?
+        {
+            Err(err) => {
+                // Getting local timelines can fail when no local tenant directory is on disk (e.g, when tenant data is being downloaded).
+                // In that case, put a warning message into log and operate normally.
+                warn!("Failed to get local timelines for tenant {tenant_id}: {err}");
+                None
+            }
+            Ok(local_timeline_infos) => Some(
+                local_timeline_infos
+                    .into_iter()
+                    .fold(0, |acc, x| acc + x.1.current_physical_size.unwrap()),
+            ),
+        };
+
     json_response(
         StatusCode::OK,
         TenantInfo {
@@ -782,6 +818,11 @@ async fn failpoints_handler(mut request: Request<Body>) -> Result<Response<Body>
 }
 
 // Run GC immediately on given timeline.
+// FIXME: This is just for tests. See test_runner/regress/test_gc.py.
+// This probably should require special authentication or a global flag to
+// enable, I don't think we want to or need to allow regular clients to invoke
+// GC.
+//     @hllinnaka in commits ec44f4b29, 3aca717f3
 #[cfg(feature = "testing")]
 async fn timeline_gc_handler(mut request: Request<Body>) -> Result<Response<Body>, ApiError> {
     let tenant_id: TenantId = parse_request_param(&request, "tenant_id")?;
@@ -789,16 +830,16 @@ async fn timeline_gc_handler(mut request: Request<Body>) -> Result<Response<Body
     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 repo = tenant_mgr::get_tenant(tenant_id, false).map_err(ApiError::NotFound)?;
     let gc_req: TimelineGcRequest = json_request(&mut request).await?;
 
     let _span_guard =
         info_span!("manual_gc", tenant = %tenant_id, timeline = %timeline_id).entered();
-    let gc_horizon = gc_req.gc_horizon.unwrap_or_else(|| tenant.get_gc_horizon());
+    let gc_horizon = gc_req.gc_horizon.unwrap_or_else(|| repo.get_gc_horizon());
 
     // Use tenant's pitr setting
-    let pitr = tenant.get_pitr_interval();
-    let result = tenant
+    let pitr = repo.get_pitr_interval();
+    let result = repo
         .gc_iteration(Some(timeline_id), gc_horizon, pitr, true)
         // FIXME: `gc_iteration` can return an error for multiple reasons; we should handle it
         // better once the types support it.
@@ -807,15 +848,19 @@ async fn timeline_gc_handler(mut request: Request<Body>) -> Result<Response<Body
 }
 
 // Run compaction immediately on given timeline.
+// FIXME This is just for tests. Don't expect this to be exposed to
+// the users or the api.
+//     @dhammika in commit a0781f229
 #[cfg(feature = "testing")]
 async fn timeline_compact_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 tenant = tenant_mgr::get_tenant(tenant_id, true).map_err(ApiError::NotFound)?;
-    let timeline = tenant
-        .get_timeline(timeline_id, true)
+    let repo = tenant_mgr::get_tenant(tenant_id, true).map_err(ApiError::NotFound)?;
+    let timeline = repo
+        .get_timeline(timeline_id)
+        .with_context(|| format!("No timeline {timeline_id} in repository for tenant {tenant_id}"))
         .map_err(ApiError::NotFound)?;
     timeline.compact().map_err(ApiError::InternalServerError)?;
 
@@ -829,9 +874,10 @@ 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 timeline = tenant
-        .get_timeline(timeline_id, true)
+    let repo = tenant_mgr::get_tenant(tenant_id, true).map_err(ApiError::NotFound)?;
+    let timeline = repo
+        .get_timeline(timeline_id)
+        .with_context(|| format!("No timeline {timeline_id} in repository for tenant {tenant_id}"))
         .map_err(ApiError::NotFound)?;
     timeline
         .checkpoint(CheckpointConfig::Forced)

pageserver/src/import_datadir.rs

@@ -12,10 +12,10 @@ use tracing::*;
 use walkdir::WalkDir;
 
 use crate::pgdatadir_mapping::*;
+use crate::reltag::{RelTag, SlruKind};
 use crate::tenant::Timeline;
 use crate::walingest::WalIngest;
 use crate::walrecord::DecodedWALRecord;
-use pageserver_api::reltag::{RelTag, SlruKind};
 use postgres_ffi::pg_constants;
 use postgres_ffi::relfile_utils::*;
 use postgres_ffi::waldecoder::WalStreamDecoder;
@@ -43,19 +43,19 @@ pub fn get_lsn_from_controlfile(path: &Path) -> Result<Lsn> {
 /// The code that deals with the checkpoint would not work right if the
 /// cluster was not shut down cleanly.
 pub fn import_timeline_from_postgres_datadir(
+    path: &Path,
     tline: &Timeline,
-    pgdata_path: &Path,
-    pgdata_lsn: Lsn,
+    lsn: Lsn,
 ) -> Result<()> {
     let mut pg_control: Option<ControlFileData> = None;
 
     // TODO this shoud be start_lsn, which is not necessarily equal to end_lsn (aka lsn)
     // Then fishing out pg_control would be unnecessary
-    let mut modification = tline.begin_modification(pgdata_lsn);
+    let mut modification = tline.begin_modification(lsn);
     modification.init_empty()?;
 
     // Import all but pg_wal
-    let all_but_wal = WalkDir::new(pgdata_path)
+    let all_but_wal = WalkDir::new(path)
         .into_iter()
         .filter_entry(|entry| !entry.path().ends_with("pg_wal"));
     for entry in all_but_wal {
@@ -63,7 +63,7 @@ pub fn import_timeline_from_postgres_datadir(
         let metadata = entry.metadata().expect("error getting dir entry metadata");
         if metadata.is_file() {
             let absolute_path = entry.path();
-            let relative_path = absolute_path.strip_prefix(pgdata_path)?;
+            let relative_path = absolute_path.strip_prefix(path)?;
 
             let file = File::open(absolute_path)?;
             let len = metadata.len() as usize;
@@ -84,7 +84,7 @@ pub fn import_timeline_from_postgres_datadir(
         "Postgres cluster was not shut down cleanly"
     );
     ensure!(
-        pg_control.checkPointCopy.redo == pgdata_lsn.0,
+        pg_control.checkPointCopy.redo == lsn.0,
         "unexpected checkpoint REDO pointer"
     );
 
@@ -92,10 +92,10 @@ pub fn import_timeline_from_postgres_datadir(
     // this reads the checkpoint record itself, advancing the tip of the timeline to
     // *after* the checkpoint record. And crucially, it initializes the 'prev_lsn'.
     import_wal(
-        &pgdata_path.join("pg_wal"),
+        &path.join("pg_wal"),
         tline,
         Lsn(pg_control.checkPointCopy.redo),
-        pgdata_lsn,
+        lsn,
     )?;
 
     Ok(())

pageserver/src/lib.rs

@@ -5,10 +5,10 @@ pub mod import_datadir;
 pub mod keyspace;
 pub mod metrics;
 pub mod page_cache;
-pub mod page_image_cache;
 pub mod page_service;
 pub mod pgdatadir_mapping;
 pub mod profiling;
+pub mod reltag;
 pub mod repository;
 pub mod storage_sync;
 pub mod task_mgr;
@@ -46,8 +46,6 @@ pub const DELTA_FILE_MAGIC: u16 = 0x5A61;
 
 pub const LOG_FILE_NAME: &str = "pageserver.log";
 
-static ZERO_PAGE: bytes::Bytes = bytes::Bytes::from_static(&[0u8; 8192]);
-
 /// Config for the Repository checkpointer
 #[derive(Debug, Clone, Copy)]
 pub enum CheckpointConfig {
@@ -121,6 +119,32 @@ impl<T> TenantTimelineValues<T> {
     fn new() -> Self {
         Self(HashMap::new())
     }
+
+    fn with_capacity(capacity: usize) -> Self {
+        Self(HashMap::with_capacity(capacity))
+    }
+
+    /// A convenience method to map certain values and omit some of them, if needed.
+    /// Tenants that won't have any timeline entries due to the filtering, will still be preserved
+    /// in the structure.
+    fn filter_map<F, NewT>(self, map: F) -> TenantTimelineValues<NewT>
+    where
+        F: Fn(T) -> Option<NewT>,
+    {
+        let capacity = self.0.len();
+        self.0.into_iter().fold(
+            TenantTimelineValues::<NewT>::with_capacity(capacity),
+            |mut new_values, (tenant_id, old_values)| {
+                let new_timeline_values = new_values.0.entry(tenant_id).or_default();
+                for (timeline_id, old_value) in old_values {
+                    if let Some(new_value) = map(old_value) {
+                        new_timeline_values.insert(timeline_id, new_value);
+                    }
+                }
+                new_values
+            },
+        )
+    }
 }
 
 /// A suffix to be used during file sync from the remote storage,
@@ -157,3 +181,35 @@ mod backoff_defaults_tests {
         );
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use crate::tenant::harness::TIMELINE_ID;
+
+    use super::*;
+
+    #[test]
+    fn tenant_timeline_value_mapping() {
+        let first_tenant = TenantId::generate();
+        let second_tenant = TenantId::generate();
+        assert_ne!(first_tenant, second_tenant);
+
+        let mut initial = TenantTimelineValues::new();
+        initial
+            .0
+            .entry(first_tenant)
+            .or_default()
+            .insert(TIMELINE_ID, "test_value");
+        let _ = initial.0.entry(second_tenant).or_default();
+        assert_eq!(initial.0.len(), 2, "Should have entries for both tenants");
+
+        let filtered = initial.filter_map(|_| None::<&str>).0;
+        assert_eq!(
+            filtered.len(),
+            2,
+            "Should have entries for both tenants even after filtering away all entries"
+        );
+        assert!(filtered.contains_key(&first_tenant));
+        assert!(filtered.contains_key(&second_tenant));
+    }
+}

pageserver/src/metrics.rs

@@ -107,20 +107,18 @@ static CURRENT_LOGICAL_SIZE: Lazy<UIntGaugeVec> = Lazy::new(|| {
 
 // Metrics for cloud upload. These metrics reflect data uploaded to cloud storage,
 // or in testing they estimate how much we would upload if we did.
-static NUM_PERSISTENT_FILES_CREATED: Lazy<IntCounterVec> = Lazy::new(|| {
-    register_int_counter_vec!(
+static NUM_PERSISTENT_FILES_CREATED: Lazy<IntCounter> = Lazy::new(|| {
+    register_int_counter!(
         "pageserver_created_persistent_files_total",
         "Number of files created that are meant to be uploaded to cloud storage",
-        &["tenant_id", "timeline_id"]
     )
     .expect("failed to define a metric")
 });
 
-static PERSISTENT_BYTES_WRITTEN: Lazy<IntCounterVec> = Lazy::new(|| {
-    register_int_counter_vec!(
+static PERSISTENT_BYTES_WRITTEN: Lazy<IntCounter> = Lazy::new(|| {
+    register_int_counter!(
         "pageserver_written_persistent_bytes_total",
         "Total bytes written that are meant to be uploaded to cloud storage",
-        &["tenant_id", "timeline_id"]
     )
     .expect("failed to define a metric")
 });
@@ -277,15 +275,11 @@ pub static TENANT_TASK_EVENTS: Lazy<IntCounterVec> = Lazy::new(|| {
 /// smallest redo processing times. These buckets allow us to measure down
 /// to 5us, which equates to 200'000 pages/sec, which equates to 1.6GB/sec.
 /// This is much better than the previous 5ms aka 200 pages/sec aka 1.6MB/sec.
-///
-/// Values up to 1s are recorded because metrics show that we have redo
-/// durations and lock times larger than 0.250s.
 macro_rules! redo_histogram_time_buckets {
     () => {
         vec![
             0.000_005, 0.000_010, 0.000_025, 0.000_050, 0.000_100, 0.000_250, 0.000_500, 0.001_000,
-            0.002_500, 0.005_000, 0.010_000, 0.025_000, 0.050_000, 0.100_000, 0.250_000, 0.500_000,
-            1.000_000,
+            0.002_500, 0.005_000, 0.010_000, 0.025_000, 0.050_000, 0.100_000, 0.250_000,
         ]
     };
 }
@@ -300,17 +294,6 @@ macro_rules! redo_histogram_count_buckets {
     };
 }
 
-macro_rules! redo_bytes_histogram_count_buckets {
-    () => {
-        // powers of (2^.5), from 2^4.5 to 2^15 (22 buckets)
-        // rounded up to the next multiple of 8 to capture any MAXALIGNed record of that size, too.
-        vec![
-            24.0, 32.0, 48.0, 64.0, 96.0, 128.0, 184.0, 256.0, 368.0, 512.0, 728.0, 1024.0, 1456.0,
-            2048.0, 2904.0, 4096.0, 5800.0, 8192.0, 11592.0, 16384.0, 23176.0, 32768.0,
-        ]
-    };
-}
-
 pub static WAL_REDO_TIME: Lazy<Histogram> = Lazy::new(|| {
     register_histogram!(
         "pageserver_wal_redo_seconds",
@@ -338,15 +321,6 @@ pub static WAL_REDO_RECORDS_HISTOGRAM: Lazy<Histogram> = Lazy::new(|| {
     .expect("failed to define a metric")
 });
 
-pub static WAL_REDO_BYTES_HISTOGRAM: Lazy<Histogram> = Lazy::new(|| {
-    register_histogram!(
-        "pageserver_wal_redo_bytes_histogram",
-        "Histogram of number of records replayed per redo",
-        redo_bytes_histogram_count_buckets!(),
-    )
-    .expect("failed to define a metric")
-});
-
 pub static WAL_REDO_RECORD_COUNTER: Lazy<IntCounter> = Lazy::new(|| {
     register_int_counter!(
         "pageserver_replayed_wal_records_total",
@@ -412,12 +386,8 @@ impl TimelineMetrics {
         let current_logical_size_gauge = CURRENT_LOGICAL_SIZE
             .get_metric_with_label_values(&[&tenant_id, &timeline_id])
             .unwrap();
-        let num_persistent_files_created = NUM_PERSISTENT_FILES_CREATED
-            .get_metric_with_label_values(&[&tenant_id, &timeline_id])
-            .unwrap();
-        let persistent_bytes_written = PERSISTENT_BYTES_WRITTEN
-            .get_metric_with_label_values(&[&tenant_id, &timeline_id])
-            .unwrap();
+        let num_persistent_files_created = NUM_PERSISTENT_FILES_CREATED.clone();
+        let persistent_bytes_written = PERSISTENT_BYTES_WRITTEN.clone();
 
         TimelineMetrics {
             tenant_id,
@@ -449,8 +419,6 @@ impl Drop for TimelineMetrics {
         let _ = WAIT_LSN_TIME.remove_label_values(&[tenant_id, timeline_id]);
         let _ = CURRENT_PHYSICAL_SIZE.remove_label_values(&[tenant_id, timeline_id]);
         let _ = CURRENT_LOGICAL_SIZE.remove_label_values(&[tenant_id, timeline_id]);
-        let _ = NUM_PERSISTENT_FILES_CREATED.remove_label_values(&[tenant_id, timeline_id]);
-        let _ = PERSISTENT_BYTES_WRITTEN.remove_label_values(&[tenant_id, timeline_id]);
 
         for op in STORAGE_TIME_OPERATIONS {
             let _ = STORAGE_TIME.remove_label_values(&[op, tenant_id, timeline_id]);

pageserver/src/page_cache.rs

@@ -108,10 +108,10 @@ enum CacheKey {
 }
 
 #[derive(Debug, PartialEq, Eq, Hash, Clone)]
-pub struct MaterializedPageHashKey {
-    pub tenant_id: TenantId,
-    pub timeline_id: TimelineId,
-    pub key: Key,
+struct MaterializedPageHashKey {
+    tenant_id: TenantId,
+    timeline_id: TimelineId,
+    key: Key,
 }
 
 #[derive(Clone)]

pageserver/src/page_image_cache.rs

@@ -1,308 +0,0 @@
-//!
-//! Global page image cache
-//!
-//! Unlike page_cache it holds only most recent version of reconstructed page images.
-//! And it uses invalidation mechanism to avoid layer ap lookups.
-
-use crate::page_cache::MaterializedPageHashKey;
-use crate::pgdatadir_mapping::{rel_block_to_key, BlockNumber};
-use crate::repository::Key;
-use crate::tenant::Timeline;
-use anyhow::{bail, Result};
-use bytes::Bytes;
-use once_cell::sync::OnceCell;
-use pageserver_api::reltag::RelTag;
-use std::collections::hash_map::DefaultHasher;
-use std::hash::{Hash, Hasher};
-use std::sync::{Arc, Condvar, Mutex};
-use utils::{
-    id::{TenantId, TimelineId},
-    lsn::Lsn,
-};
-
-static PAGE_CACHE: OnceCell<Mutex<PageImageCache>> = OnceCell::new();
-const TEST_PAGE_CACHE_SIZE: usize = 50;
-
-enum PageImageState {
-    Vacant,                        // entry is not used
-    Loaded(Option<Bytes>),         // page is loaded or has failed
-    Loading(Option<Arc<Condvar>>), // page in process of loading, Condvar is created on demand when some thread need to wait load completion
-}
-
-struct CacheEntry {
-    key: MaterializedPageHashKey,
-
-    // next+prev are used for LRU L2-list and next is also used for L1 free pages list
-    next: usize,
-    prev: usize,
-
-    collision: usize, // L1 hash collision chain
-
-    state: PageImageState,
-}
-
-pub struct PageImageCache {
-    free_list: usize, // L1 list of free entries
-    pages: Vec<CacheEntry>,
-    hash_table: Vec<usize>, // indexes in pages array
-}
-
-///
-/// Initialize the page cache. This must be called once at page server startup.
-///
-pub fn init(size: usize) {
-    if PAGE_CACHE
-        .set(Mutex::new(PageImageCache::new(size)))
-        .is_err()
-    {
-        panic!("page cache already initialized");
-    }
-}
-
-///
-/// Get a handle to the page cache.
-///
-pub fn get() -> &'static Mutex<PageImageCache> {
-    //
-    // In unit tests, page server startup doesn't happen and no one calls
-    // page_image_cache::init(). Initialize it here with a tiny cache, so that the
-    // page cache is usable in unit tests.
-    //
-    if cfg!(test) {
-        PAGE_CACHE.get_or_init(|| Mutex::new(PageImageCache::new(TEST_PAGE_CACHE_SIZE)))
-    } else {
-        PAGE_CACHE.get().expect("page cache not initialized")
-    }
-}
-
-fn hash<T: Hash>(t: &T) -> usize {
-    let mut s = DefaultHasher::new();
-    t.hash(&mut s);
-    s.finish() as usize
-}
-
-impl PageImageCache {
-    fn new(size: usize) -> Self {
-        let mut pages: Vec<CacheEntry> = Vec::with_capacity(size + 1);
-        let hash_table = vec![0usize; size];
-
-        // Dummy key
-        let dummy_key = MaterializedPageHashKey {
-            key: Key::MIN,
-            tenant_id: TenantId::from([0u8; 16]),
-            timeline_id: TimelineId::from([0u8; 16]),
-        };
-
-        // LRU list head
-        pages.push(CacheEntry {
-            key: dummy_key.clone(),
-            next: 0,
-            prev: 0,
-            collision: 0,
-            state: PageImageState::Vacant,
-        });
-
-        // Construct L1 free page list
-        for i in 0..size {
-            pages.push(CacheEntry {
-                key: dummy_key.clone(),
-                next: i + 2, // build L1-list of free pages
-                prev: 0,
-                collision: 0,
-                state: PageImageState::Vacant,
-            });
-        }
-        pages[size - 1].next = 0; // en of free page list
-
-        PageImageCache {
-            free_list: 1,
-            pages,
-            hash_table,
-        }
-    }
-
-    // Unlink from L2-list
-    fn unlink(&mut self, index: usize) {
-        let next = self.pages[index].next;
-        let prev = self.pages[index].prev;
-        self.pages[next].prev = prev;
-        self.pages[prev].next = next;
-    }
-
-    // Link in L2-list after specified element
-    fn link_after(&mut self, after: usize, index: usize) {
-        let next = self.pages[after].next;
-        self.pages[index].prev = after;
-        self.pages[index].next = next;
-        self.pages[next].prev = index;
-        self.pages[after].next = index;
-    }
-
-    fn prune(&mut self, index: usize) {
-        self.pages[index].prev = index;
-        self.pages[index].next = index;
-    }
-
-    fn is_empty(&self, index: usize) -> bool {
-        self.pages[index].next == index
-    }
-}
-
-// Remove entry from cache: o page invalidation or drop relation
-pub fn remove(key: Key, tenant_id: TenantId, timeline_id: TimelineId) {
-    let key = MaterializedPageHashKey {
-        key,
-        tenant_id,
-        timeline_id,
-    };
-    let this = get();
-    let mut cache = this.lock().unwrap();
-    let h = hash(&key) % cache.hash_table.len();
-    let mut index = cache.hash_table[h];
-    let mut prev = 0usize;
-    while index != 0 {
-        if cache.pages[index].key == key {
-            if !cache.is_empty(index) {
-                cache.pages[index].state = PageImageState::Vacant;
-                // Remove from LRU list
-                cache.unlink(index);
-                // Insert entry in free list
-                cache.pages[index].next = cache.free_list;
-                cache.free_list = index;
-            } else {
-                // Page is process of loading: we can not remove it righ now,
-                // so just mark for deletion
-                cache.pages[index].next = 0; // make is_empty == false
-            }
-            // Remove from hash table
-            if prev == 0 {
-                cache.hash_table[h] = cache.pages[index].collision;
-            } else {
-                cache.pages[prev].collision = cache.pages[index].collision;
-            }
-            break;
-        }
-        prev = index;
-        index = cache.pages[index].collision;
-    }
-    // It's Ok if image not found
-}
-
-// Find or load page image in the cache
-pub fn lookup(timeline: &Timeline, rel: RelTag, blkno: BlockNumber, lsn: Lsn) -> Result<Bytes> {
-    let key = MaterializedPageHashKey {
-        key: rel_block_to_key(rel, blkno),
-        tenant_id: timeline.tenant_id,
-        timeline_id: timeline.timeline_id,
-    };
-    let this = get();
-    let mut cache = this.lock().unwrap();
-    let h = hash(&key) % cache.hash_table.len();
-
-    'lookup: loop {
-        let mut index = cache.hash_table[h];
-        while index != 0 {
-            if cache.pages[index].key == key {
-                // cache hit
-                match &cache.pages[index].state {
-                    PageImageState::Loaded(cached_page) => {
-                        // Move to the head of LRU list
-                        let page = cached_page.clone();
-                        cache.unlink(index);
-                        cache.link_after(0, index);
-                        return page.ok_or_else(|| anyhow::anyhow!("page loading failed earlier"));
-                    }
-                    PageImageState::Loading(event) => {
-                        // Create event on which to sleep if not yet assigned
-                        let cv = match event {
-                            None => {
-                                let cv = Arc::new(Condvar::new());
-                                cache.pages[index].state =
-                                    PageImageState::Loading(Some(cv.clone()));
-                                cv
-                            }
-                            Some(cv) => cv.clone(),
-                        };
-                        cache = cv.wait(cache).unwrap();
-                        // Retry lookup
-                        continue 'lookup;
-                    }
-                    PageImageState::Vacant => bail!("Vacant entry is not expected here"),
-                };
-            }
-            index = cache.pages[index].collision;
-        }
-        // Cache miss
-        index = cache.free_list;
-        if index == 0 {
-            // no free items
-            let victim = cache.pages[0].prev; // take least recently used element from the tail of LRU list
-            assert!(victim != 0);
-            // Remove victim from hash table
-            let h = hash(&cache.pages[victim].key) % cache.hash_table.len();
-            index = cache.hash_table[h];
-            let mut prev = 0usize;
-            while index != victim {
-                assert!(index != 0);
-                prev = index;
-                index = cache.pages[index].collision;
-            }
-            if prev == 0 {
-                cache.hash_table[h] = cache.pages[victim].collision;
-            } else {
-                cache.pages[prev].collision = cache.pages[victim].collision;
-            }
-            // and from LRU list
-            cache.unlink(victim);
-
-            index = victim;
-        } else {
-            // Use next free item
-            cache.free_list = cache.pages[index].next;
-        }
-        // Make is_empty(index) == true. If entry is removed in process of loaded,
-        // it will be updated so that !is_empty(index)
-        cache.prune(index);
-
-        // Insert in hash table
-        cache.pages[index].collision = cache.hash_table[h];
-        cache.hash_table[h] = index;
-
-        cache.pages[index].key = key;
-        cache.pages[index].state = PageImageState::Loading(None);
-        drop(cache); //release lock
-
-        // Load page
-        let res = timeline.get_rel_page_at_lsn(rel, blkno, lsn, true);
-
-        cache = this.lock().unwrap();
-        if let PageImageState::Loading(event) = &cache.pages[index].state {
-            // Are there soMe waiting threads?
-            if let Some(cv) = event {
-                // If so, then wakeup them
-                cv.notify_all();
-            }
-        } else {
-            bail!("Loading state is expected");
-        }
-        if cache.is_empty(index) {
-            // entry was not marked as deleted {
-            // Page is loaded
-
-            // match &res { ... } is same as `res.as_ref().ok().cloned()`
-            cache.pages[index].state = PageImageState::Loaded(match &res {
-                Ok(page) => Some(page.clone()),
-                Err(_) => None,
-            });
-            // Link the page to the head of LRU list
-            cache.link_after(0, index);
-        } else {
-            cache.pages[index].state = PageImageState::Vacant;
-            // Return page to free list
-            cache.pages[index].next = cache.free_list;
-            cache.free_list = index;
-        }
-        // only the first one gets the full error from `get_rel_page_at_lsn`
-        return res;
-    }
-}

pageserver/src/page_service.rs

@@ -10,15 +10,8 @@
 //
 
 use anyhow::{bail, ensure, Context, Result};
-use bytes::Bytes;
+use bytes::{Buf, BufMut, Bytes, BytesMut};
 use futures::{Stream, StreamExt};
-use pageserver_api::models::{
-    PagestreamBeMessage, PagestreamDbSizeRequest, PagestreamDbSizeResponse,
-    PagestreamErrorResponse, PagestreamExistsRequest, PagestreamExistsResponse,
-    PagestreamFeMessage, PagestreamGetPageRequest, PagestreamGetPageResponse,
-    PagestreamNblocksRequest, PagestreamNblocksResponse,
-};
-
 use std::io;
 use std::net::TcpListener;
 use std::str;
@@ -39,10 +32,10 @@ use utils::{
 
 use crate::basebackup;
 use crate::config::{PageServerConf, ProfilingConfig};
-use crate::import_datadir::import_wal_from_tar;
+use crate::import_datadir::{import_basebackup_from_tar, import_wal_from_tar};
 use crate::metrics::{LIVE_CONNECTIONS_COUNT, SMGR_QUERY_TIME};
-use crate::page_image_cache;
 use crate::profiling::profpoint_start;
+use crate::reltag::RelTag;
 use crate::task_mgr;
 use crate::task_mgr::TaskKind;
 use crate::tenant::Timeline;
@@ -52,6 +45,163 @@ use crate::CheckpointConfig;
 use postgres_ffi::pg_constants::DEFAULTTABLESPACE_OID;
 use postgres_ffi::BLCKSZ;
 
+// Wrapped in libpq CopyData
+enum PagestreamFeMessage {
+    Exists(PagestreamExistsRequest),
+    Nblocks(PagestreamNblocksRequest),
+    GetPage(PagestreamGetPageRequest),
+    DbSize(PagestreamDbSizeRequest),
+}
+
+// Wrapped in libpq CopyData
+enum PagestreamBeMessage {
+    Exists(PagestreamExistsResponse),
+    Nblocks(PagestreamNblocksResponse),
+    GetPage(PagestreamGetPageResponse),
+    Error(PagestreamErrorResponse),
+    DbSize(PagestreamDbSizeResponse),
+}
+
+#[derive(Debug)]
+struct PagestreamExistsRequest {
+    latest: bool,
+    lsn: Lsn,
+    rel: RelTag,
+}
+
+#[derive(Debug)]
+struct PagestreamNblocksRequest {
+    latest: bool,
+    lsn: Lsn,
+    rel: RelTag,
+}
+
+#[derive(Debug)]
+struct PagestreamGetPageRequest {
+    latest: bool,
+    lsn: Lsn,
+    rel: RelTag,
+    blkno: u32,
+}
+
+#[derive(Debug)]
+struct PagestreamDbSizeRequest {
+    latest: bool,
+    lsn: Lsn,
+    dbnode: u32,
+}
+
+#[derive(Debug)]
+struct PagestreamExistsResponse {
+    exists: bool,
+}
+
+#[derive(Debug)]
+struct PagestreamNblocksResponse {
+    n_blocks: u32,
+}
+
+#[derive(Debug)]
+struct PagestreamGetPageResponse {
+    page: Bytes,
+}
+
+#[derive(Debug)]
+struct PagestreamErrorResponse {
+    message: String,
+}
+
+#[derive(Debug)]
+struct PagestreamDbSizeResponse {
+    db_size: i64,
+}
+
+impl PagestreamFeMessage {
+    fn parse(mut body: Bytes) -> anyhow::Result<PagestreamFeMessage> {
+        // TODO these gets can fail
+
+        // these correspond to the NeonMessageTag enum in pagestore_client.h
+        //
+        // TODO: consider using protobuf or serde bincode for less error prone
+        // serialization.
+        let msg_tag = body.get_u8();
+        match msg_tag {
+            0 => Ok(PagestreamFeMessage::Exists(PagestreamExistsRequest {
+                latest: body.get_u8() != 0,
+                lsn: Lsn::from(body.get_u64()),
+                rel: RelTag {
+                    spcnode: body.get_u32(),
+                    dbnode: body.get_u32(),
+                    relnode: body.get_u32(),
+                    forknum: body.get_u8(),
+                },
+            })),
+            1 => Ok(PagestreamFeMessage::Nblocks(PagestreamNblocksRequest {
+                latest: body.get_u8() != 0,
+                lsn: Lsn::from(body.get_u64()),
+                rel: RelTag {
+                    spcnode: body.get_u32(),
+                    dbnode: body.get_u32(),
+                    relnode: body.get_u32(),
+                    forknum: body.get_u8(),
+                },
+            })),
+            2 => Ok(PagestreamFeMessage::GetPage(PagestreamGetPageRequest {
+                latest: body.get_u8() != 0,
+                lsn: Lsn::from(body.get_u64()),
+                rel: RelTag {
+                    spcnode: body.get_u32(),
+                    dbnode: body.get_u32(),
+                    relnode: body.get_u32(),
+                    forknum: body.get_u8(),
+                },
+                blkno: body.get_u32(),
+            })),
+            3 => Ok(PagestreamFeMessage::DbSize(PagestreamDbSizeRequest {
+                latest: body.get_u8() != 0,
+                lsn: Lsn::from(body.get_u64()),
+                dbnode: body.get_u32(),
+            })),
+            _ => bail!("unknown smgr message tag: {},'{:?}'", msg_tag, body),
+        }
+    }
+}
+
+impl PagestreamBeMessage {
+    fn serialize(&self) -> Bytes {
+        let mut bytes = BytesMut::new();
+
+        match self {
+            Self::Exists(resp) => {
+                bytes.put_u8(100); /* tag from pagestore_client.h */
+                bytes.put_u8(resp.exists as u8);
+            }
+
+            Self::Nblocks(resp) => {
+                bytes.put_u8(101); /* tag from pagestore_client.h */
+                bytes.put_u32(resp.n_blocks);
+            }
+
+            Self::GetPage(resp) => {
+                bytes.put_u8(102); /* tag from pagestore_client.h */
+                bytes.put(&resp.page[..]);
+            }
+
+            Self::Error(resp) => {
+                bytes.put_u8(103); /* tag from pagestore_client.h */
+                bytes.put(resp.message.as_bytes());
+                bytes.put_u8(0); // null terminator
+            }
+            Self::DbSize(resp) => {
+                bytes.put_u8(104); /* tag from pagestore_client.h */
+                bytes.put_i64(resp.db_size);
+            }
+        }
+
+        bytes.into()
+    }
+}
+
 fn copyin_stream(pgb: &mut PostgresBackend) -> impl Stream<Item = io::Result<Bytes>> + '_ {
     async_stream::try_stream! {
         loop {
@@ -350,8 +500,11 @@ impl PageServerHandler {
         task_mgr::associate_with(Some(tenant_id), Some(timeline_id));
         // Create empty timeline
         info!("creating new timeline");
-        let tenant = tenant_mgr::get_tenant(tenant_id, true)?;
-        let timeline = tenant.create_empty_timeline(timeline_id, base_lsn, pg_version)?;
+        let timeline = tenant_mgr::get_tenant(tenant_id, true)?.create_empty_timeline(
+            timeline_id,
+            base_lsn,
+            pg_version,
+        )?;
 
         // TODO mark timeline as not ready until it reaches end_lsn.
         // We might have some wal to import as well, and we should prevent compute
@@ -374,8 +527,7 @@ impl PageServerHandler {
         // - use block_in_place()
         let mut copyin_stream = Box::pin(copyin_stream(pgb));
         let reader = SyncIoBridge::new(StreamReader::new(&mut copyin_stream));
-        tokio::task::block_in_place(|| timeline.import_basebackup_from_tar(reader, base_lsn))?;
-        timeline.initialize()?;
+        tokio::task::block_in_place(|| import_basebackup_from_tar(&timeline, reader, base_lsn))?;
 
         // Drain the rest of the Copy data
         let mut bytes_after_tar = 0;
@@ -392,6 +544,12 @@ impl PageServerHandler {
         // It wouldn't work if base came from vanilla postgres though,
         // since we discard some log files.
 
+        // Flush data to disk, then upload to s3
+        info!("flushing layers");
+        timeline.checkpoint(CheckpointConfig::Flush)?;
+
+        timeline.launch_wal_receiver()?;
+
         info!("done");
         Ok(())
     }
@@ -582,12 +740,8 @@ impl PageServerHandler {
         // current profiling is based on a thread-local variable, so it doesn't work
         // across awaits
         let _profiling_guard = profpoint_start(self.conf, ProfilingConfig::PageRequests);
+        let page = timeline.get_rel_page_at_lsn(req.rel, req.blkno, lsn, req.latest)?;
 
-        let page = if req.latest {
-            page_image_cache::lookup(timeline, req.rel, req.blkno, lsn)
-        } else {
-            timeline.get_rel_page_at_lsn(req.rel, req.blkno, lsn, false)
-        }?;
         Ok(PagestreamBeMessage::GetPage(PagestreamGetPageResponse {
             page,
         }))
@@ -914,8 +1068,7 @@ impl postgres_backend_async::Handler for PageServerHandler {
 }
 
 fn get_local_timeline(tenant_id: TenantId, timeline_id: TimelineId) -> Result<Arc<Timeline>> {
-    tenant_mgr::get_tenant(tenant_id, true)
-        .and_then(|tenant| tenant.get_timeline(timeline_id, true))
+    tenant_mgr::get_tenant(tenant_id, true).and_then(|tenant| tenant.get_timeline(timeline_id))
 }
 
 ///

pageserver/src/pgdatadir_mapping.rs

@@ -7,12 +7,12 @@
 //! Clarify that)
 //!
 use crate::keyspace::{KeySpace, KeySpaceAccum};
+use crate::reltag::{RelTag, SlruKind};
 use crate::repository::*;
 use crate::tenant::Timeline;
 use crate::walrecord::NeonWalRecord;
 use anyhow::{bail, ensure, Result};
 use bytes::{Buf, Bytes};
-use pageserver_api::reltag::{RelTag, SlruKind};
 use postgres_ffi::relfile_utils::{FSM_FORKNUM, VISIBILITYMAP_FORKNUM};
 use postgres_ffi::BLCKSZ;
 use postgres_ffi::{Oid, TimestampTz, TransactionId};
@@ -1179,7 +1179,7 @@ fn rel_dir_to_key(spcnode: Oid, dbnode: Oid) -> Key {
     }
 }
 
-pub fn rel_block_to_key(rel: RelTag, blknum: BlockNumber) -> Key {
+fn rel_block_to_key(rel: RelTag, blknum: BlockNumber) -> Key {
     Key {
         field1: 0x00,
         field2: rel.spcnode,
@@ -1373,17 +1373,6 @@ fn is_rel_block_key(key: Key) -> bool {
     key.field1 == 0x00 && key.field4 != 0
 }
 
-pub fn is_rel_fsm_block_key(key: Key) -> bool {
-    key.field1 == 0x00 && key.field4 != 0 && key.field5 == FSM_FORKNUM && key.field6 != 0xffffffff
-}
-
-pub fn is_rel_vm_block_key(key: Key) -> bool {
-    key.field1 == 0x00
-        && key.field4 != 0
-        && key.field5 == VISIBILITYMAP_FORKNUM
-        && key.field6 != 0xffffffff
-}
-
 pub fn key_to_slru_block(key: Key) -> Result<(SlruKind, u32, BlockNumber)> {
     Ok(match key.field1 {
         0x01 => {
@@ -1414,9 +1403,7 @@ pub fn create_test_timeline(
     timeline_id: utils::id::TimelineId,
     pg_version: u32,
 ) -> Result<std::sync::Arc<Timeline>> {
-    let tline = tenant
-        .create_empty_timeline(timeline_id, Lsn(8), pg_version)?
-        .initialize()?;
+    let tline = tenant.create_empty_timeline(timeline_id, Lsn(8), pg_version)?;
     let mut m = tline.begin_modification(Lsn(8));
     m.init_empty()?;
     m.commit()?;

pageserver/src/storage_sync.rs

@@ -169,14 +169,9 @@ use self::{
     upload::{upload_index_part, upload_timeline_layers, UploadedTimeline},
 };
 use crate::{
-    config::PageServerConf,
-    exponential_backoff,
-    storage_sync::index::{LayerFileMetadata, RemoteIndex},
-    task_mgr,
-    task_mgr::TaskKind,
-    task_mgr::BACKGROUND_RUNTIME,
-    tenant::metadata::TimelineMetadata,
-    tenant_mgr::{attach_local_tenants, TenantAttachData},
+    config::PageServerConf, exponential_backoff, storage_sync::index::RemoteIndex, task_mgr,
+    task_mgr::TaskKind, task_mgr::BACKGROUND_RUNTIME, tenant::metadata::TimelineMetadata,
+    tenant_mgr::attach_local_tenants,
 };
 use crate::{
     metrics::{IMAGE_SYNC_TIME, REMAINING_SYNC_ITEMS, REMOTE_INDEX_UPLOAD},
@@ -193,7 +188,7 @@ static SYNC_QUEUE: OnceCell<SyncQueue> = OnceCell::new();
 
 /// A timeline status to share with pageserver's sync counterpart,
 /// after comparing local and remote timeline state.
-#[derive(Clone, PartialEq, Eq)]
+#[derive(Clone)]
 pub enum LocalTimelineInitStatus {
     /// The timeline has every remote layer present locally.
     /// There could be some layers requiring uploading,
@@ -316,7 +311,7 @@ impl SyncQueue {
 
 /// A task to run in the async download/upload loop.
 /// Limited by the number of retries, after certain threshold the failing task gets evicted and the timeline disabled.
-#[derive(Debug, Clone, PartialEq, Eq)]
+#[derive(Debug, Clone)]
 enum SyncTask {
     /// A checkpoint outcome with possible local file updates that need actualization in the remote storage.
     /// Not necessary more fresh than the one already uploaded.
@@ -427,7 +422,7 @@ impl SyncTaskBatch {
                             .extend(new_delete.data.deleted_layers.iter().cloned());
                     }
                     if let Some(batch_upload) = &mut self.upload {
-                        let not_deleted = |layer: &PathBuf, _: &mut LayerFileMetadata| {
+                        let not_deleted = |layer: &PathBuf| {
                             !new_delete.data.layers_to_delete.contains(layer)
                                 && !new_delete.data.deleted_layers.contains(layer)
                         };
@@ -455,35 +450,21 @@ impl SyncTaskBatch {
 #[derive(Debug, Clone, PartialEq, Eq)]
 struct LayersUpload {
     /// Layer file path in the pageserver workdir, that were added for the corresponding checkpoint.
-    layers_to_upload: HashMap<PathBuf, LayerFileMetadata>,
+    layers_to_upload: HashSet<PathBuf>,
     /// Already uploaded layers. Used to store the data about the uploads between task retries
     /// and to record the data into the remote index after the task got completed or evicted.
-    uploaded_layers: HashMap<PathBuf, LayerFileMetadata>,
+    uploaded_layers: HashSet<PathBuf>,
     metadata: Option<TimelineMetadata>,
 }
 
 /// A timeline download task.
 /// Does not contain the file list to download, to allow other
 /// parts of the pageserer code to schedule the task
-/// without using the remote index or any other ways to list the remote timeline files.
+/// without using the remote index or any other ways to list the remote timleine files.
 /// Skips the files that are already downloaded.
 #[derive(Debug, Clone, PartialEq, Eq)]
 struct LayersDownload {
     layers_to_skip: HashSet<PathBuf>,
-
-    /// Paths which have been downloaded, and had their metadata verified or generated.
-    ///
-    /// Metadata generation happens when upgrading from past version of `IndexPart`.
-    gathered_metadata: HashMap<PathBuf, LayerFileMetadata>,
-}
-
-impl LayersDownload {
-    fn from_skipped_layers(layers_to_skip: HashSet<PathBuf>) -> Self {
-        LayersDownload {
-            layers_to_skip,
-            gathered_metadata: HashMap::default(),
-        }
-    }
 }
 
 #[derive(Debug, Clone, PartialEq, Eq)]
@@ -505,7 +486,7 @@ struct LayersDeletion {
 pub fn schedule_layer_upload(
     tenant_id: TenantId,
     timeline_id: TimelineId,
-    layers_to_upload: HashMap<PathBuf, LayerFileMetadata>,
+    layers_to_upload: HashSet<PathBuf>,
     metadata: Option<TimelineMetadata>,
 ) {
     let sync_queue = match SYNC_QUEUE.get() {
@@ -522,7 +503,7 @@ pub fn schedule_layer_upload(
         },
         SyncTask::upload(LayersUpload {
             layers_to_upload,
-            uploaded_layers: HashMap::new(),
+            uploaded_layers: HashSet::new(),
             metadata,
         }),
     );
@@ -580,44 +561,18 @@ pub fn schedule_layer_download(tenant_id: TenantId, timeline_id: TimelineId) {
             tenant_id,
             timeline_id,
         },
-        SyncTask::download(LayersDownload::from_skipped_layers(HashSet::new())),
+        SyncTask::download(LayersDownload {
+            layers_to_skip: HashSet::new(),
+        }),
     );
     debug!("Download task for tenant {tenant_id}, timeline {timeline_id} sent")
 }
 
-/// Local existing timeline files
-///
-/// Values of this type serve different meanings in different contexts. On startup, collected
-/// timelines come with the full collected information and when signalling readyness to attach
-/// after completed download. After the download the file information is no longer carried, because
-/// it is already merged into [`RemoteTimeline`].
-#[derive(Debug)]
-pub struct TimelineLocalFiles(TimelineMetadata, HashMap<PathBuf, LayerFileMetadata>);
-
-impl TimelineLocalFiles {
-    pub fn metadata(&self) -> &TimelineMetadata {
-        &self.0
-    }
-
-    /// Called during startup, for all of the local files with full metadata.
-    pub(crate) fn collected(
-        metadata: TimelineMetadata,
-        timeline_files: HashMap<PathBuf, LayerFileMetadata>,
-    ) -> TimelineLocalFiles {
-        TimelineLocalFiles(metadata, timeline_files)
-    }
-
-    /// Called near the end of tenant initialization, to signal readyness to attach tenants.
-    pub(crate) fn ready(metadata: TimelineMetadata) -> Self {
-        TimelineLocalFiles(metadata, HashMap::new())
-    }
-}
-
 /// Launch a thread to perform remote storage sync tasks.
 /// See module docs for loop step description.
 pub fn spawn_storage_sync_task(
     conf: &'static PageServerConf,
-    local_timeline_files: HashMap<TenantId, HashMap<TimelineId, TimelineLocalFiles>>,
+    local_timeline_files: TenantTimelineValues<(TimelineMetadata, HashSet<PathBuf>)>,
     storage: GenericRemoteStorage,
     max_concurrent_timelines_sync: NonZeroUsize,
     max_sync_errors: NonZeroU32,
@@ -640,7 +595,7 @@ pub fn spawn_storage_sync_task(
     let mut keys_for_index_part_downloads = HashSet::new();
     let mut timelines_to_sync = HashMap::new();
 
-    for (tenant_id, timeline_data) in local_timeline_files {
+    for (tenant_id, timeline_data) in local_timeline_files.0 {
         if timeline_data.is_empty() {
             info!("got empty tenant {}", tenant_id);
             let _ = empty_tenants.0.entry(tenant_id).or_default();
@@ -743,7 +698,7 @@ async fn storage_sync_loop(
                         "Sync loop step completed, {} new tenant state update(s)",
                         updated_tenants.len()
                     );
-                    let mut timelines_to_attach = HashMap::new();
+                    let mut timelines_to_attach = TenantTimelineValues::new();
                     let index_accessor = index.read().await;
                     for tenant_id in updated_tenants {
                         let tenant_entry = match index_accessor.tenant_entry(&tenant_id) {
@@ -769,16 +724,12 @@ async fn storage_sync_loop(
                             // and register them all at once in a tenant for download
                             // to be submitted in a single operation to tenant
                             // so it can apply them at once to internal timeline map.
-                            timelines_to_attach.insert(
+                            timelines_to_attach.0.insert(
                                 tenant_id,
-                                TenantAttachData::Ready(
-                                    tenant_entry
-                                        .iter()
-                                        .map(|(&id, entry)| {
-                                            (id, TimelineLocalFiles::ready(entry.metadata.clone()))
-                                        })
-                                        .collect(),
-                                ),
+                                tenant_entry
+                                    .iter()
+                                    .map(|(&id, entry)| (id, entry.metadata.clone()))
+                                    .collect(),
                             );
                         }
                     }
@@ -1020,27 +971,15 @@ async fn download_timeline_data(
         }
         DownloadedTimeline::Successful(mut download_data) => {
             match update_local_metadata(conf, sync_id, current_remote_timeline).await {
-                Ok(()) => {
-                    let mut g = index.write().await;
-
-                    match g.set_awaits_download(&sync_id, false) {
-                        Ok(()) => {
-                            let timeline = g
-                                .timeline_entry_mut(&sync_id)
-                                .expect("set_awaits_download verified existence");
-
-                            timeline.merge_metadata_from_downloaded(
-                                &download_data.data.gathered_metadata,
-                            );
-
-                            register_sync_status(sync_id, sync_start, TASK_NAME, Some(true));
-                            return DownloadStatus::Downloaded;
-                        }
-                        Err(e) => {
-                            error!("Timeline {sync_id} was expected to be in the remote index after a successful download, but it's absent: {e:?}");
-                        }
-                    };
-                }
+                Ok(()) => match index.write().await.set_awaits_download(&sync_id, false) {
+                    Ok(()) => {
+                        register_sync_status(sync_id, sync_start, TASK_NAME, Some(true));
+                        return DownloadStatus::Downloaded;
+                    }
+                    Err(e) => {
+                        error!("Timeline {sync_id} was expected to be in the remote index after a successful download, but it's absent: {e:?}");
+                    }
+                },
                 Err(e) => {
                     error!("Failed to update local timeline metadata: {e:?}");
                     download_data.retries += 1;
@@ -1243,18 +1182,11 @@ async fn update_remote_data(
                         }
                         if upload_failed {
                             existing_entry.add_upload_failures(
-                                uploaded_data
-                                    .layers_to_upload
-                                    .iter()
-                                    .map(|(k, v)| (k.to_owned(), v.to_owned())),
+                                uploaded_data.layers_to_upload.iter().cloned(),
                             );
                         } else {
-                            existing_entry.add_timeline_layers(
-                                uploaded_data
-                                    .uploaded_layers
-                                    .iter()
-                                    .map(|(k, v)| (k.to_owned(), v.to_owned())),
-                            );
+                            existing_entry
+                                .add_timeline_layers(uploaded_data.uploaded_layers.iter().cloned());
                         }
                     }
                     RemoteDataUpdate::Delete(layers_to_remove) => {
@@ -1274,19 +1206,11 @@ async fn update_remote_data(
                     };
                     let mut new_remote_timeline = RemoteTimeline::new(new_metadata.clone());
                     if upload_failed {
-                        new_remote_timeline.add_upload_failures(
-                            uploaded_data
-                                .layers_to_upload
-                                .iter()
-                                .map(|(k, v)| (k.to_owned(), v.to_owned())),
-                        );
+                        new_remote_timeline
+                            .add_upload_failures(uploaded_data.layers_to_upload.iter().cloned());
                     } else {
-                        new_remote_timeline.add_timeline_layers(
-                            uploaded_data
-                                .uploaded_layers
-                                .iter()
-                                .map(|(k, v)| (k.to_owned(), v.to_owned())),
-                        );
+                        new_remote_timeline
+                            .add_timeline_layers(uploaded_data.uploaded_layers.iter().cloned());
                     }
 
                     index_accessor.add_timeline_entry(sync_id, new_remote_timeline.clone());
@@ -1334,14 +1258,13 @@ async fn validate_task_retries(
 fn schedule_first_sync_tasks(
     index: &mut RemoteTimelineIndex,
     sync_queue: &SyncQueue,
-    local_timeline_files: HashMap<TenantTimelineId, TimelineLocalFiles>,
+    local_timeline_files: HashMap<TenantTimelineId, (TimelineMetadata, HashSet<PathBuf>)>,
 ) -> TenantTimelineValues<LocalTimelineInitStatus> {
     let mut local_timeline_init_statuses = TenantTimelineValues::new();
 
     let mut new_sync_tasks = VecDeque::with_capacity(local_timeline_files.len());
 
-    for (sync_id, local_timeline) in local_timeline_files {
-        let TimelineLocalFiles(local_metadata, local_files) = local_timeline;
+    for (sync_id, (local_metadata, local_files)) in local_timeline_files {
         match index.timeline_entry_mut(&sync_id) {
             Some(remote_timeline) => {
                 let (timeline_status, awaits_download) = compare_local_and_remote_timeline(
@@ -1385,7 +1308,7 @@ fn schedule_first_sync_tasks(
                     sync_id,
                     SyncTask::upload(LayersUpload {
                         layers_to_upload: local_files,
-                        uploaded_layers: HashMap::new(),
+                        uploaded_layers: HashSet::new(),
                         metadata: Some(local_metadata.clone()),
                     }),
                 ));
@@ -1412,46 +1335,20 @@ fn compare_local_and_remote_timeline(
     new_sync_tasks: &mut VecDeque<(TenantTimelineId, SyncTask)>,
     sync_id: TenantTimelineId,
     local_metadata: TimelineMetadata,
-    local_files: HashMap<PathBuf, LayerFileMetadata>,
+    local_files: HashSet<PathBuf>,
     remote_entry: &RemoteTimeline,
 ) -> (LocalTimelineInitStatus, bool) {
     let _entered = info_span!("compare_local_and_remote_timeline", sync_id = %sync_id).entered();
 
-    let needed_to_download_files = remote_entry
-        .stored_files()
-        .iter()
-        .filter_map(|(layer_file, remote_metadata)| {
-            if let Some(local_metadata) = local_files.get(layer_file) {
-                match (remote_metadata.file_size(), local_metadata.file_size()) {
-                    (Some(x), Some(y)) if x == y => { None },
-                    (None, Some(_)) => {
-                        // upgrading from an earlier IndexPart without metadata
-                        None
-                    },
-                    _ => {
-                        // having to deal with other than (Some(x), Some(y)) where x != y here is a
-                        // bummer, but see #2582 and #2610 for attempts and discussion.
-                        warn!("Redownloading locally existing {layer_file:?} due to size mismatch, size on index: {:?}, on disk: {:?}", remote_metadata.file_size(), local_metadata.file_size());
-                        Some(layer_file)
-                    },
-                }
-            } else {
-                // doesn't exist locally
-                Some(layer_file)
-            }
-        })
-        .collect::<HashSet<_>>();
+    let remote_files = remote_entry.stored_files();
 
-    let (initial_timeline_status, awaits_download) = if !needed_to_download_files.is_empty() {
+    let number_of_layers_to_download = remote_files.difference(&local_files).count();
+    let (initial_timeline_status, awaits_download) = if number_of_layers_to_download > 0 {
         new_sync_tasks.push_back((
             sync_id,
-            SyncTask::download(LayersDownload::from_skipped_layers(
-                local_files
-                    .keys()
-                    .filter(|path| !needed_to_download_files.contains(path))
-                    .cloned()
-                    .collect(),
-            )),
+            SyncTask::download(LayersDownload {
+                layers_to_skip: local_files.clone(),
+            }),
         ));
         info!("NeedsSync");
         (LocalTimelineInitStatus::NeedsSync, true)
@@ -1466,22 +1363,15 @@ fn compare_local_and_remote_timeline(
     };
 
     let layers_to_upload = local_files
-        .iter()
-        .filter_map(|(local_file, metadata)| {
-            if !remote_entry.stored_files().contains_key(local_file) {
-                Some((local_file.to_owned(), metadata.to_owned()))
-            } else {
-                None
-            }
-        })
-        .collect::<HashMap<_, _>>();
-
+        .difference(remote_files)
+        .cloned()
+        .collect::<HashSet<_>>();
     if !layers_to_upload.is_empty() {
         new_sync_tasks.push_back((
             sync_id,
             SyncTask::upload(LayersUpload {
                 layers_to_upload,
-                uploaded_layers: HashMap::new(),
+                uploaded_layers: HashSet::new(),
                 metadata: Some(local_metadata),
             }),
         ));
@@ -1537,12 +1427,11 @@ mod test_utils {
         let timeline_path = harness.timeline_path(&timeline_id);
         fs::create_dir_all(&timeline_path).await?;
 
-        let mut layers_to_upload = HashMap::with_capacity(filenames.len());
+        let mut layers_to_upload = HashSet::with_capacity(filenames.len());
         for &file in filenames {
             let file_path = timeline_path.join(file);
             fs::write(&file_path, dummy_contents(file).into_bytes()).await?;
-            let metadata = LayerFileMetadata::new(file_path.metadata()?.len());
-            layers_to_upload.insert(file_path, metadata);
+            layers_to_upload.insert(file_path);
         }
 
         fs::write(
@@ -1553,7 +1442,7 @@ mod test_utils {
 
         Ok(LayersUpload {
             layers_to_upload,
-            uploaded_layers: HashMap::new(),
+            uploaded_layers: HashSet::new(),
             metadata: Some(metadata),
         })
     }
@@ -1608,13 +1497,12 @@ mod tests {
         assert!(sync_id_2 != sync_id_3);
         assert!(sync_id_3 != TEST_SYNC_ID);
 
-        let download_task =
-            SyncTask::download(LayersDownload::from_skipped_layers(HashSet::from([
-                PathBuf::from("sk"),
-            ])));
+        let download_task = SyncTask::download(LayersDownload {
+            layers_to_skip: HashSet::from([PathBuf::from("sk")]),
+        });
         let upload_task = SyncTask::upload(LayersUpload {
-            layers_to_upload: HashMap::from([(PathBuf::from("up"), LayerFileMetadata::new(123))]),
-            uploaded_layers: HashMap::from([(PathBuf::from("upl"), LayerFileMetadata::new(123))]),
+            layers_to_upload: HashSet::from([PathBuf::from("up")]),
+            uploaded_layers: HashSet::from([PathBuf::from("upl")]),
             metadata: Some(dummy_metadata(Lsn(2))),
         });
         let delete_task = SyncTask::delete(LayersDeletion {
@@ -1658,10 +1546,12 @@ mod tests {
         let sync_queue = SyncQueue::new(NonZeroUsize::new(100).unwrap());
         assert_eq!(sync_queue.len(), 0);
 
-        let download = LayersDownload::from_skipped_layers(HashSet::from([PathBuf::from("sk")]));
+        let download = LayersDownload {
+            layers_to_skip: HashSet::from([PathBuf::from("sk")]),
+        };
         let upload = LayersUpload {
-            layers_to_upload: HashMap::from([(PathBuf::from("up"), LayerFileMetadata::new(123))]),
-            uploaded_layers: HashMap::from([(PathBuf::from("upl"), LayerFileMetadata::new(123))]),
+            layers_to_upload: HashSet::from([PathBuf::from("up")]),
+            uploaded_layers: HashSet::from([PathBuf::from("upl")]),
             metadata: Some(dummy_metadata(Lsn(2))),
         };
         let delete = LayersDeletion {
@@ -1709,10 +1599,18 @@ mod tests {
     #[tokio::test]
     async fn same_task_id_same_tasks_batch() {
         let sync_queue = SyncQueue::new(NonZeroUsize::new(1).unwrap());
-        let download_1 = LayersDownload::from_skipped_layers(HashSet::from([PathBuf::from("sk1")]));
-        let download_2 = LayersDownload::from_skipped_layers(HashSet::from([PathBuf::from("sk2")]));
-        let download_3 = LayersDownload::from_skipped_layers(HashSet::from([PathBuf::from("sk3")]));
-        let download_4 = LayersDownload::from_skipped_layers(HashSet::from([PathBuf::from("sk4")]));
+        let download_1 = LayersDownload {
+            layers_to_skip: HashSet::from([PathBuf::from("sk1")]),
+        };
+        let download_2 = LayersDownload {
+            layers_to_skip: HashSet::from([PathBuf::from("sk2")]),
+        };
+        let download_3 = LayersDownload {
+            layers_to_skip: HashSet::from([PathBuf::from("sk3")]),
+        };
+        let download_4 = LayersDownload {
+            layers_to_skip: HashSet::from([PathBuf::from("sk4")]),
+        };
 
         let sync_id_2 = TenantTimelineId {
             tenant_id: TenantId::from_array(hex!("22223344556677881122334455667788")),
@@ -1736,15 +1634,15 @@ mod tests {
             Some(SyncTaskBatch {
                 download: Some(SyncData {
                     retries: 0,
-                    data: LayersDownload::from_skipped_layers(
-                        {
+                    data: LayersDownload {
+                        layers_to_skip: {
                             let mut set = HashSet::new();
                             set.extend(download_1.layers_to_skip.into_iter());
                             set.extend(download_2.layers_to_skip.into_iter());
                             set.extend(download_4.layers_to_skip.into_iter());
                             set
                         },
-                    )
+                    }
                 }),
                 upload: None,
                 delete: None,
@@ -1760,148 +1658,4 @@ mod tests {
             "Should have one task left out of the batch"
         );
     }
-
-    mod local_and_remote_comparisons {
-        use super::*;
-
-        #[test]
-        fn ready() {
-            let mut new_sync_tasks = VecDeque::default();
-            let sync_id = TenantTimelineId::generate();
-            let local_metadata = dummy_metadata(0x02.into());
-            let local_files =
-                HashMap::from([(PathBuf::from("first_file"), LayerFileMetadata::new(123))]);
-            let mut remote_entry = RemoteTimeline::new(local_metadata.clone());
-            remote_entry
-                .add_timeline_layers([(PathBuf::from("first_file"), LayerFileMetadata::new(123))]);
-
-            let (status, sync_needed) = compare_local_and_remote_timeline(
-                &mut new_sync_tasks,
-                sync_id,
-                local_metadata.clone(),
-                local_files,
-                &remote_entry,
-            );
-
-            assert_eq!(
-                status,
-                LocalTimelineInitStatus::LocallyComplete(local_metadata)
-            );
-            assert!(!sync_needed);
-
-            assert!(new_sync_tasks.is_empty(), "{:?}", new_sync_tasks);
-        }
-
-        #[test]
-        fn needs_download() {
-            let mut new_sync_tasks = VecDeque::default();
-            let sync_id = TenantTimelineId::generate();
-            let local_metadata = dummy_metadata(0x02.into());
-            let local_files = HashMap::default();
-            let mut remote_entry = RemoteTimeline::new(local_metadata.clone());
-            remote_entry
-                .add_timeline_layers([(PathBuf::from("first_file"), LayerFileMetadata::new(123))]);
-
-            let (status, sync_needed) = compare_local_and_remote_timeline(
-                &mut new_sync_tasks,
-                sync_id,
-                local_metadata,
-                local_files.clone(),
-                &remote_entry,
-            );
-
-            assert_eq!(status, LocalTimelineInitStatus::NeedsSync);
-            assert!(sync_needed);
-
-            let new_sync_tasks = new_sync_tasks.into_iter().collect::<Vec<_>>();
-
-            assert_eq!(
-                &new_sync_tasks,
-                &[(
-                    sync_id,
-                    SyncTask::download(LayersDownload::from_skipped_layers(
-                        local_files.keys().cloned().collect()
-                    ))
-                )]
-            );
-        }
-
-        #[test]
-        fn redownload_is_not_needed_on_upgrade() {
-            // originally the implementation missed the `(None, Some(_))` case in the match, and
-            // proceeded to always redownload if the remote metadata was not available.
-
-            let mut new_sync_tasks = VecDeque::default();
-            let sync_id = TenantTimelineId::generate();
-
-            let local_metadata = dummy_metadata(0x02.into());
-
-            // type system would in general allow that LayerFileMetadata would be created with
-            // file_size: None, however `LayerFileMetadata::default` is only allowed from tests,
-            // and so everywhere within the system valid LayerFileMetadata is being created, it is
-            // created through `::new`.
-            let local_files =
-                HashMap::from([(PathBuf::from("first_file"), LayerFileMetadata::new(123))]);
-
-            let mut remote_entry = RemoteTimeline::new(local_metadata.clone());
-
-            // RemoteTimeline is constructed out of an older version IndexPart, which didn't carry
-            // any metadata.
-            remote_entry
-                .add_timeline_layers([(PathBuf::from("first_file"), LayerFileMetadata::default())]);
-
-            let (status, sync_needed) = compare_local_and_remote_timeline(
-                &mut new_sync_tasks,
-                sync_id,
-                local_metadata.clone(),
-                local_files,
-                &remote_entry,
-            );
-
-            assert_eq!(
-                status,
-                LocalTimelineInitStatus::LocallyComplete(local_metadata)
-            );
-            assert!(!sync_needed);
-        }
-
-        #[test]
-        fn needs_upload() {
-            let mut new_sync_tasks = VecDeque::default();
-            let sync_id = TenantTimelineId::generate();
-            let local_metadata = dummy_metadata(0x02.into());
-            let local_files =
-                HashMap::from([(PathBuf::from("first_file"), LayerFileMetadata::new(123))]);
-            let mut remote_entry = RemoteTimeline::new(local_metadata.clone());
-            remote_entry.add_timeline_layers([]);
-
-            let (status, sync_needed) = compare_local_and_remote_timeline(
-                &mut new_sync_tasks,
-                sync_id,
-                local_metadata.clone(),
-                local_files.clone(),
-                &remote_entry,
-            );
-
-            assert_eq!(
-                status,
-                LocalTimelineInitStatus::LocallyComplete(local_metadata.clone())
-            );
-            assert!(!sync_needed);
-
-            let new_sync_tasks = new_sync_tasks.into_iter().collect::<Vec<_>>();
-
-            assert_eq!(
-                &new_sync_tasks,
-                &[(
-                    sync_id,
-                    SyncTask::upload(LayersUpload {
-                        layers_to_upload: local_files,
-                        uploaded_layers: HashMap::default(),
-                        metadata: Some(local_metadata),
-                    })
-                )]
-            );
-        }
-    }
 }

pageserver/src/storage_sync/delete.rs

@@ -171,7 +171,7 @@ mod tests {
         let local_timeline_path = harness.timeline_path(&TIMELINE_ID);
         let timeline_upload =
             create_local_timeline(&harness, TIMELINE_ID, &layer_files, metadata.clone()).await?;
-        for (local_path, _metadata) in timeline_upload.layers_to_upload {
+        for local_path in timeline_upload.layers_to_upload {
             let remote_path =
                 local_storage.resolve_in_storage(&local_storage.remote_object_id(&local_path)?)?;
             let remote_parent_dir = remote_path.parent().unwrap();

pageserver/src/storage_sync/download.rs

@@ -16,13 +16,9 @@ use tokio::{
 };
 use tracing::{debug, error, info, warn};
 
-use crate::{
-    config::PageServerConf,
-    storage_sync::{index::LayerFileMetadata, SyncTask},
-    TEMP_FILE_SUFFIX,
-};
+use crate::{config::PageServerConf, storage_sync::SyncTask, TEMP_FILE_SUFFIX};
 use utils::{
-    crashsafe::path_with_suffix_extension,
+    crashsafe_dir::path_with_suffix_extension,
     id::{TenantId, TenantTimelineId, TimelineId},
 };
 
@@ -223,14 +219,8 @@ pub(super) async fn download_timeline_layers<'a>(
 
     let layers_to_download = remote_timeline
         .stored_files()
-        .iter()
-        .filter_map(|(layer_path, metadata)| {
-            if !download.layers_to_skip.contains(layer_path) {
-                Some((layer_path.to_owned(), metadata.to_owned()))
-            } else {
-                None
-            }
-        })
+        .difference(&download.layers_to_skip)
+        .cloned()
         .collect::<Vec<_>>();
 
     debug!("Layers to download: {layers_to_download:?}");
@@ -243,129 +233,89 @@ pub(super) async fn download_timeline_layers<'a>(
 
     let mut download_tasks = layers_to_download
         .into_iter()
-        .map(|(layer_destination_path, metadata)| async move {
+        .map(|layer_destination_path| async move {
+            if layer_destination_path.exists() {
+                debug!(
+                    "Layer already exists locally, skipping download: {}",
+                    layer_destination_path.display()
+                );
+            } else {
+                // Perform a rename inspired by durable_rename from file_utils.c.
+                // The sequence:
+                //     write(tmp)
+                //     fsync(tmp)
+                //     rename(tmp, new)
+                //     fsync(new)
+                //     fsync(parent)
+                // For more context about durable_rename check this email from postgres mailing list:
+                // https://www.postgresql.org/message-id/56583BDD.9060302@2ndquadrant.com
+                // If pageserver crashes the temp file will be deleted on startup and re-downloaded.
+                let temp_file_path =
+                    path_with_suffix_extension(&layer_destination_path, TEMP_FILE_SUFFIX);
 
-            match layer_destination_path.metadata() {
-                Ok(m) if m.is_file() => {
-                    // the file exists from earlier round when we failed after renaming it as
-                    // layer_destination_path
-                    let verified = if let Some(expected) = metadata.file_size() {
-                        m.len() == expected
-                    } else {
-                        // behaviour before recording metadata was to accept any existing
-                        true
-                    };
+                let mut destination_file =
+                    fs::File::create(&temp_file_path).await.with_context(|| {
+                        format!(
+                            "Failed to create a destination file for layer '{}'",
+                            temp_file_path.display()
+                        )
+                    })?;
 
-                    if verified {
-                        debug!(
-                            "Layer already exists locally, skipping download: {}",
-                            layer_destination_path.display()
-                        );
-                        return Ok((layer_destination_path, LayerFileMetadata::new(m.len())))
-                    } else {
-                        // no need to remove it, it will be overwritten by fs::rename
-                        // after successful download
-                        warn!("Downloaded layer exists already but layer file metadata mismatches: {}, metadata {:?}", layer_destination_path.display(), metadata);
-                    }
-                }
-                Ok(m) => {
-                    return Err(anyhow::anyhow!("Downloaded layer destination exists but is not a file: {m:?}, target needs to be removed/archived manually: {layer_destination_path:?}"));
-                }
-                Err(_) => {
-                    // behave as the file didn't exist
-                }
+                let mut layer_download = storage.download_storage_object(None, &layer_destination_path)
+                    .await
+                    .with_context(|| {
+                        format!(
+                            "Failed to initiate the download the layer for {sync_id} into file '{}'",
+                            temp_file_path.display()
+                        )
+                    })?;
+                io::copy(&mut layer_download.download_stream, &mut destination_file)
+                    .await
+                    .with_context(|| {
+                        format!(
+                            "Failed to download the layer for {sync_id} into file '{}'",
+                            temp_file_path.display()
+                        )
+                    })?;
+
+                // Tokio doc here: https://docs.rs/tokio/1.17.0/tokio/fs/struct.File.html states that:
+                // A file will not be closed immediately when it goes out of scope if there are any IO operations
+                // that have not yet completed. To ensure that a file is closed immediately when it is dropped,
+                // you should call flush before dropping it.
+                //
+                // From the tokio code I see that it waits for pending operations to complete. There shouldn't be any because
+                // we assume that `destination_file` file is fully written. I.e there is no pending .write(...).await operations.
+                // But for additional safety let's check/wait for any pending operations.
+                destination_file.flush().await.with_context(|| {
+                    format!(
+                        "failed to flush source file at {}",
+                        temp_file_path.display()
+                    )
+                })?;
+
+                // not using sync_data because it can lose file size update
+                destination_file.sync_all().await.with_context(|| {
+                    format!(
+                        "failed to fsync source file at {}",
+                        temp_file_path.display()
+                    )
+                })?;
+                drop(destination_file);
+
+                fail::fail_point!("remote-storage-download-pre-rename", |_| {
+                    anyhow::bail!("remote-storage-download-pre-rename failpoint triggered")
+                });
+
+                fs::rename(&temp_file_path, &layer_destination_path).await?;
+
+                fsync_path(&layer_destination_path).await.with_context(|| {
+                    format!(
+                        "Cannot fsync layer destination path {}",
+                        layer_destination_path.display(),
+                    )
+                })?;
             }
-
-            // Perform a rename inspired by durable_rename from file_utils.c.
-            // The sequence:
-            //     write(tmp)
-            //     fsync(tmp)
-            //     rename(tmp, new)
-            //     fsync(new)
-            //     fsync(parent)
-            // For more context about durable_rename check this email from postgres mailing list:
-            // https://www.postgresql.org/message-id/56583BDD.9060302@2ndquadrant.com
-            // If pageserver crashes the temp file will be deleted on startup and re-downloaded.
-            let temp_file_path =
-                path_with_suffix_extension(&layer_destination_path, TEMP_FILE_SUFFIX);
-
-            // TODO: this doesn't use the cached fd for some reason?
-            let mut destination_file =
-                fs::File::create(&temp_file_path).await.with_context(|| {
-                    format!(
-                        "Failed to create a destination file for layer '{}'",
-                        temp_file_path.display()
-                    )
-                })?;
-
-            let mut layer_download = storage.download_storage_object(None, &layer_destination_path)
-                .await
-                .with_context(|| {
-                    format!(
-                        "Failed to initiate the download the layer for {sync_id} into file '{}'",
-                        temp_file_path.display()
-                    )
-                })?;
-
-            let bytes_amount = io::copy(&mut layer_download.download_stream, &mut destination_file)
-                .await
-                .with_context(|| {
-                    format!(
-                        "Failed to download the layer for {sync_id} into file '{}'",
-                        temp_file_path.display()
-                    )
-                })?;
-
-            // Tokio doc here: https://docs.rs/tokio/1.17.0/tokio/fs/struct.File.html states that:
-            // A file will not be closed immediately when it goes out of scope if there are any IO operations
-            // that have not yet completed. To ensure that a file is closed immediately when it is dropped,
-            // you should call flush before dropping it.
-            //
-            // From the tokio code I see that it waits for pending operations to complete. There shouldn't be any because
-            // we assume that `destination_file` file is fully written. I.e there is no pending .write(...).await operations.
-            // But for additional safety let's check/wait for any pending operations.
-            destination_file.flush().await.with_context(|| {
-                format!(
-                    "failed to flush source file at {}",
-                    temp_file_path.display()
-                )
-            })?;
-
-            match metadata.file_size() {
-                Some(expected) if expected != bytes_amount => {
-                    anyhow::bail!(
-                        "According to layer file metadata should had downloaded {expected} bytes but downloaded {bytes_amount} bytes into file '{}'",
-                        temp_file_path.display()
-                    );
-                },
-                Some(_) | None => {
-                    // matches, or upgrading from an earlier IndexPart version
-                }
-            }
-
-            // not using sync_data because it can lose file size update
-            destination_file.sync_all().await.with_context(|| {
-                format!(
-                    "failed to fsync source file at {}",
-                    temp_file_path.display()
-                )
-            })?;
-            drop(destination_file);
-
-            fail::fail_point!("remote-storage-download-pre-rename", |_| {
-                anyhow::bail!("remote-storage-download-pre-rename failpoint triggered")
-            });
-
-            fs::rename(&temp_file_path, &layer_destination_path).await?;
-
-            fsync_path(&layer_destination_path).await.with_context(|| {
-                format!(
-                    "Cannot fsync layer destination path {}",
-                    layer_destination_path.display(),
-                )
-            })?;
-
-            Ok::<_, anyhow::Error>((layer_destination_path, LayerFileMetadata::new(bytes_amount)))
+            Ok::<_, anyhow::Error>(layer_destination_path)
         })
         .collect::<FuturesUnordered<_>>();
 
@@ -374,12 +324,9 @@ pub(super) async fn download_timeline_layers<'a>(
     let mut undo = HashSet::new();
     while let Some(download_result) = download_tasks.next().await {
         match download_result {
-            Ok((downloaded_path, metadata)) => {
+            Ok(downloaded_path) => {
                 undo.insert(downloaded_path.clone());
-                download.layers_to_skip.insert(downloaded_path.clone());
-                // what if the key existed already? ignore, because then we would had
-                // downloaded a partial file, and had to retry
-                download.gathered_metadata.insert(downloaded_path, metadata);
+                download.layers_to_skip.insert(downloaded_path);
             }
             Err(e) => {
                 errors_happened = true;
@@ -402,8 +349,6 @@ pub(super) async fn download_timeline_layers<'a>(
         );
         for item in undo {
             download.layers_to_skip.remove(&item);
-            // intentionally don't clear the gathered_metadata because it exists for fsync_path
-            // failure on parent directory
         }
         errors_happened = true;
     }
@@ -508,9 +453,9 @@ mod tests {
         let timeline_upload =
             create_local_timeline(&harness, TIMELINE_ID, &layer_files, metadata.clone()).await?;
 
-        for local_path in timeline_upload.layers_to_upload.keys() {
+        for local_path in timeline_upload.layers_to_upload {
             let remote_path =
-                local_storage.resolve_in_storage(&storage.remote_object_id(local_path)?)?;
+                local_storage.resolve_in_storage(&storage.remote_object_id(&local_path)?)?;
             let remote_parent_dir = remote_path.parent().unwrap();
             if !remote_parent_dir.exists() {
                 fs::create_dir_all(&remote_parent_dir).await?;
@@ -528,19 +473,11 @@ mod tests {
 
         let mut remote_timeline = RemoteTimeline::new(metadata.clone());
         remote_timeline.awaits_download = true;
-        remote_timeline.add_timeline_layers(layer_files.iter().map(|layer| {
-            let layer_path = local_timeline_path.join(layer);
-
-            // this could had also been LayerFileMetadata::default(), but since in this test we
-            // don't do the merge operation done by storage_sync::download_timeline_data, it would
-            // not be merged back to timeline.
-            let metadata_from_upload = timeline_upload
-                .layers_to_upload
-                .get(&layer_path)
-                .expect("layer must exist in previously uploaded paths")
-                .to_owned();
-            (layer_path, metadata_from_upload)
-        }));
+        remote_timeline.add_timeline_layers(
+            layer_files
+                .iter()
+                .map(|layer| local_timeline_path.join(layer)),
+        );
 
         let download_data = match download_timeline_layers(
             harness.conf,
@@ -550,9 +487,9 @@ mod tests {
             sync_id,
             SyncData::new(
                 current_retries,
-                LayersDownload::from_skipped_layers(HashSet::from([
-                    local_timeline_path.join("layer_to_skip")
-                ])),
+                LayersDownload {
+                    layers_to_skip: HashSet::from([local_timeline_path.join("layer_to_skip")]),
+                },
             ),
         )
         .await
@@ -615,7 +552,12 @@ mod tests {
             &sync_queue,
             None,
             sync_id,
-            SyncData::new(0, LayersDownload::from_skipped_layers(HashSet::new())),
+            SyncData::new(
+                0,
+                LayersDownload {
+                    layers_to_skip: HashSet::new(),
+                },
+            ),
         )
         .await;
         assert!(
@@ -634,7 +576,12 @@ mod tests {
             &sync_queue,
             Some(&not_expecting_download_remote_timeline),
             sync_id,
-            SyncData::new(0, LayersDownload::from_skipped_layers(HashSet::new())),
+            SyncData::new(
+                0,
+                LayersDownload {
+                    layers_to_skip: HashSet::new(),
+                },
+            ),
         )
         .await;
         assert!(

pageserver/src/storage_sync/index.rs

@@ -212,8 +212,8 @@ impl RemoteTimelineIndex {
 /// Restored index part data about the timeline, stored in the remote index.
 #[derive(Debug, Clone)]
 pub struct RemoteTimeline {
-    timeline_layers: HashMap<PathBuf, LayerFileMetadata>,
-    missing_layers: HashMap<PathBuf, LayerFileMetadata>,
+    timeline_layers: HashSet<PathBuf>,
+    missing_layers: HashSet<PathBuf>,
 
     pub metadata: TimelineMetadata,
     pub awaits_download: bool,
@@ -222,161 +222,62 @@ pub struct RemoteTimeline {
 impl RemoteTimeline {
     pub fn new(metadata: TimelineMetadata) -> Self {
         Self {
-            timeline_layers: HashMap::default(),
-            missing_layers: HashMap::default(),
+            timeline_layers: HashSet::new(),
+            missing_layers: HashSet::new(),
             metadata,
             awaits_download: false,
         }
     }
 
-    pub fn add_timeline_layers(
-        &mut self,
-        new_layers: impl IntoIterator<Item = (PathBuf, LayerFileMetadata)>,
-    ) {
-        self.timeline_layers.extend(new_layers);
+    pub fn add_timeline_layers(&mut self, new_layers: impl IntoIterator<Item = PathBuf>) {
+        self.timeline_layers.extend(new_layers.into_iter());
     }
 
-    pub fn add_upload_failures(
-        &mut self,
-        upload_failures: impl IntoIterator<Item = (PathBuf, LayerFileMetadata)>,
-    ) {
-        self.missing_layers.extend(upload_failures);
+    pub fn add_upload_failures(&mut self, upload_failures: impl IntoIterator<Item = PathBuf>) {
+        self.missing_layers.extend(upload_failures.into_iter());
     }
 
     pub fn remove_layers(&mut self, layers_to_remove: &HashSet<PathBuf>) {
         self.timeline_layers
-            .retain(|layer, _| !layers_to_remove.contains(layer));
+            .retain(|layer| !layers_to_remove.contains(layer));
         self.missing_layers
-            .retain(|layer, _| !layers_to_remove.contains(layer));
+            .retain(|layer| !layers_to_remove.contains(layer));
     }
 
     /// Lists all layer files in the given remote timeline. Omits the metadata file.
-    pub fn stored_files(&self) -> &HashMap<PathBuf, LayerFileMetadata> {
+    pub fn stored_files(&self) -> &HashSet<PathBuf> {
         &self.timeline_layers
     }
 
-    /// Combines metadata gathered or verified during downloading needed layer files to metadata on
-    /// the [`RemoteIndex`], so it can be uploaded later.
-    pub fn merge_metadata_from_downloaded(
-        &mut self,
-        downloaded: &HashMap<PathBuf, LayerFileMetadata>,
-    ) {
-        downloaded.iter().for_each(|(path, metadata)| {
-            if let Some(upgraded) = self.timeline_layers.get_mut(path) {
-                upgraded.merge(metadata);
-            }
-        });
-    }
-
     pub fn from_index_part(timeline_path: &Path, index_part: IndexPart) -> anyhow::Result<Self> {
         let metadata = TimelineMetadata::from_bytes(&index_part.metadata_bytes)?;
-        let default_metadata = &IndexLayerMetadata::default();
-
-        let find_metadata = |key: &RelativePath| -> LayerFileMetadata {
-            index_part
-                .layer_metadata
-                .get(key)
-                .unwrap_or(default_metadata)
-                .into()
-        };
-
         Ok(Self {
-            timeline_layers: index_part
-                .timeline_layers
-                .iter()
-                .map(|layer_path| (layer_path.as_path(timeline_path), find_metadata(layer_path)))
-                .collect(),
-            missing_layers: index_part
-                .missing_layers
-                .iter()
-                .map(|layer_path| (layer_path.as_path(timeline_path), find_metadata(layer_path)))
-                .collect(),
+            timeline_layers: to_local_paths(timeline_path, index_part.timeline_layers),
+            missing_layers: to_local_paths(timeline_path, index_part.missing_layers),
             metadata,
             awaits_download: false,
         })
     }
 }
 
-/// Metadata gathered for each of the layer files.
-///
-/// Fields have to be `Option`s because remote [`IndexPart`]'s can be from different version, which
-/// might have less or more metadata depending if upgrading or rolling back an upgrade.
-#[derive(Debug, Clone, PartialEq, Eq, PartialOrd, Ord)]
-#[cfg_attr(test, derive(Default))]
-pub struct LayerFileMetadata {
-    file_size: Option<u64>,
-}
-
-impl From<&'_ IndexLayerMetadata> for LayerFileMetadata {
-    fn from(other: &IndexLayerMetadata) -> Self {
-        LayerFileMetadata {
-            file_size: other.file_size,
-        }
-    }
-}
-
-impl LayerFileMetadata {
-    pub fn new(file_size: u64) -> Self {
-        LayerFileMetadata {
-            file_size: Some(file_size),
-        }
-    }
-
-    pub fn file_size(&self) -> Option<u64> {
-        self.file_size
-    }
-
-    /// Metadata has holes due to version upgrades. This method is called to upgrade self with the
-    /// other value.
-    ///
-    /// This is called on the possibly outdated version.
-    pub fn merge(&mut self, other: &Self) {
-        self.file_size = other.file_size.or(self.file_size);
-    }
-}
-
 /// Part of the remote index, corresponding to a certain timeline.
 /// Contains the data about all files in the timeline, present remotely and its metadata.
-///
-/// This type needs to be backwards and forwards compatible. When changing the fields,
-/// remember to add a test case for the changed version.
 #[serde_as]
 #[derive(Debug, PartialEq, Eq, Clone, Serialize, Deserialize)]
 pub struct IndexPart {
-    /// Debugging aid describing the version of this type.
-    #[serde(default)]
-    version: usize,
-
-    /// Each of the layers present on remote storage.
-    ///
-    /// Additional metadata can might exist in `layer_metadata`.
     timeline_layers: HashSet<RelativePath>,
-
     /// Currently is not really used in pageserver,
     /// present to manually keep track of the layer files that pageserver might never retrieve.
     ///
     /// Such "holes" might appear if any upload task was evicted on an error threshold:
     /// the this layer will only be rescheduled for upload on pageserver restart.
     missing_layers: HashSet<RelativePath>,
-
-    /// Per layer file metadata, which can be present for a present or missing layer file.
-    ///
-    /// Older versions of `IndexPart` will not have this property or have only a part of metadata
-    /// that latest version stores.
-    #[serde(default)]
-    layer_metadata: HashMap<RelativePath, IndexLayerMetadata>,
-
     #[serde_as(as = "DisplayFromStr")]
     disk_consistent_lsn: Lsn,
     metadata_bytes: Vec<u8>,
 }
 
 impl IndexPart {
-    /// When adding or modifying any parts of `IndexPart`, increment the version so that it can be
-    /// used to understand later versions.
-    ///
-    /// Version is currently informative only.
-    const LATEST_VERSION: usize = 1;
     pub const FILE_NAME: &'static str = "index_part.json";
 
     #[cfg(test)]
@@ -387,10 +288,8 @@ impl IndexPart {
         metadata_bytes: Vec<u8>,
     ) -> Self {
         Self {
-            version: Self::LATEST_VERSION,
             timeline_layers,
             missing_layers,
-            layer_metadata: HashMap::default(),
             disk_consistent_lsn,
             metadata_bytes,
         }
@@ -405,68 +304,35 @@ impl IndexPart {
         remote_timeline: RemoteTimeline,
     ) -> anyhow::Result<Self> {
         let metadata_bytes = remote_timeline.metadata.to_bytes()?;
-
-        let mut layer_metadata = HashMap::new();
-
-        let mut missing_layers = HashSet::new();
-
-        separate_paths_and_metadata(
-            timeline_path,
-            &remote_timeline.missing_layers,
-            &mut missing_layers,
-            &mut layer_metadata,
-        )
-        .context("Failed to convert missing layers' paths to relative ones")?;
-
-        let mut timeline_layers = HashSet::new();
-
-        separate_paths_and_metadata(
-            timeline_path,
-            &remote_timeline.timeline_layers,
-            &mut timeline_layers,
-            &mut layer_metadata,
-        )
-        .context("Failed to convert timeline layers' paths to relative ones")?;
-
         Ok(Self {
-            version: Self::LATEST_VERSION,
-            timeline_layers,
-            missing_layers,
-            layer_metadata,
+            timeline_layers: to_relative_paths(timeline_path, remote_timeline.timeline_layers)
+                .context("Failed to convert timeline layers' paths to relative ones")?,
+            missing_layers: to_relative_paths(timeline_path, remote_timeline.missing_layers)
+                .context("Failed to convert missing layers' paths to relative ones")?,
             disk_consistent_lsn: remote_timeline.metadata.disk_consistent_lsn(),
             metadata_bytes,
         })
     }
 }
 
-/// Serialized form of [`LayerFileMetadata`].
-#[derive(Debug, PartialEq, Eq, Clone, Serialize, Deserialize, Default)]
-pub struct IndexLayerMetadata {
-    file_size: Option<u64>,
-}
-
-impl From<&'_ LayerFileMetadata> for IndexLayerMetadata {
-    fn from(other: &'_ LayerFileMetadata) -> Self {
-        IndexLayerMetadata {
-            file_size: other.file_size,
-        }
-    }
-}
-
-fn separate_paths_and_metadata(
+fn to_local_paths(
     timeline_path: &Path,
-    input: &HashMap<PathBuf, LayerFileMetadata>,
-    output: &mut HashSet<RelativePath>,
-    layer_metadata: &mut HashMap<RelativePath, IndexLayerMetadata>,
-) -> anyhow::Result<()> {
-    for (path, metadata) in input {
-        let rel_path = RelativePath::new(timeline_path, path)?;
-        let metadata = IndexLayerMetadata::from(metadata);
+    paths: impl IntoIterator<Item = RelativePath>,
+) -> HashSet<PathBuf> {
+    paths
+        .into_iter()
+        .map(|path| path.as_path(timeline_path))
+        .collect()
+}
 
-        layer_metadata.insert(rel_path.clone(), metadata);
-        output.insert(rel_path);
-    }
-    Ok(())
+fn to_relative_paths(
+    timeline_path: &Path,
+    paths: impl IntoIterator<Item = PathBuf>,
+) -> anyhow::Result<HashSet<RelativePath>> {
+    paths
+        .into_iter()
+        .map(|path| RelativePath::new(timeline_path, path))
+        .collect()
 }
 
 #[cfg(test)]
@@ -491,13 +357,13 @@ mod tests {
             DEFAULT_PG_VERSION,
         );
         let remote_timeline = RemoteTimeline {
-            timeline_layers: HashMap::from([
-                (timeline_path.join("layer_1"), LayerFileMetadata::new(1)),
-                (timeline_path.join("layer_2"), LayerFileMetadata::new(2)),
+            timeline_layers: HashSet::from([
+                timeline_path.join("layer_1"),
+                timeline_path.join("layer_2"),
             ]),
-            missing_layers: HashMap::from([
-                (timeline_path.join("missing_1"), LayerFileMetadata::new(3)),
-                (timeline_path.join("missing_2"), LayerFileMetadata::new(4)),
+            missing_layers: HashSet::from([
+                timeline_path.join("missing_1"),
+                timeline_path.join("missing_2"),
             ]),
             metadata: metadata.clone(),
             awaits_download: false,
@@ -619,13 +485,13 @@ mod tests {
         let conversion_result = IndexPart::from_remote_timeline(
             &timeline_path,
             RemoteTimeline {
-                timeline_layers: HashMap::from([
-                    (PathBuf::from("bad_path"), LayerFileMetadata::new(1)),
-                    (timeline_path.join("layer_2"), LayerFileMetadata::new(2)),
+                timeline_layers: HashSet::from([
+                    PathBuf::from("bad_path"),
+                    timeline_path.join("layer_2"),
                 ]),
-                missing_layers: HashMap::from([
-                    (timeline_path.join("missing_1"), LayerFileMetadata::new(3)),
-                    (timeline_path.join("missing_2"), LayerFileMetadata::new(4)),
+                missing_layers: HashSet::from([
+                    timeline_path.join("missing_1"),
+                    timeline_path.join("missing_2"),
                 ]),
                 metadata: metadata.clone(),
                 awaits_download: false,
@@ -636,13 +502,13 @@ mod tests {
         let conversion_result = IndexPart::from_remote_timeline(
             &timeline_path,
             RemoteTimeline {
-                timeline_layers: HashMap::from([
-                    (timeline_path.join("layer_1"), LayerFileMetadata::new(1)),
-                    (timeline_path.join("layer_2"), LayerFileMetadata::new(2)),
+                timeline_layers: HashSet::from([
+                    timeline_path.join("layer_1"),
+                    timeline_path.join("layer_2"),
                 ]),
-                missing_layers: HashMap::from([
-                    (PathBuf::from("bad_path"), LayerFileMetadata::new(3)),
-                    (timeline_path.join("missing_2"), LayerFileMetadata::new(4)),
+                missing_layers: HashSet::from([
+                    PathBuf::from("bad_path"),
+                    timeline_path.join("missing_2"),
                 ]),
                 metadata,
                 awaits_download: false,
@@ -650,63 +516,4 @@ mod tests {
         );
         assert!(conversion_result.is_err(), "Should not be able to convert metadata with missing layer paths that are not in the timeline directory");
     }
-
-    #[test]
-    fn v0_indexpart_is_parsed() {
-        let example = r#"{
-            "timeline_layers":["000000000000000000000000000000000000-FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF__0000000001696070-00000000016960E9"],
-            "missing_layers":["not_a_real_layer_but_adding_coverage"],
-            "disk_consistent_lsn":"0/16960E8",
-            "metadata_bytes":[113,11,159,210,0,54,0,4,0,0,0,0,1,105,96,232,1,0,0,0,0,1,105,96,112,0,0,0,0,0,0,0,0,0,0,0,0,0,1,105,96,112,0,0,0,0,1,105,96,112,0,0,0,14,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0]
-        }"#;
-
-        let expected = IndexPart {
-            version: 0,
-            timeline_layers: [RelativePath("000000000000000000000000000000000000-FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF__0000000001696070-00000000016960E9".to_owned())].into_iter().collect(),
-            missing_layers: [RelativePath("not_a_real_layer_but_adding_coverage".to_owned())].into_iter().collect(),
-            layer_metadata: HashMap::default(),
-            disk_consistent_lsn: "0/16960E8".parse::<Lsn>().unwrap(),
-            metadata_bytes: [113,11,159,210,0,54,0,4,0,0,0,0,1,105,96,232,1,0,0,0,0,1,105,96,112,0,0,0,0,0,0,0,0,0,0,0,0,0,1,105,96,112,0,0,0,0,1,105,96,112,0,0,0,14,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0].to_vec(),
-        };
-
-        let part = serde_json::from_str::<IndexPart>(example).unwrap();
-        assert_eq!(part, expected);
-    }
-
-    #[test]
-    fn v1_indexpart_is_parsed() {
-        let example = r#"{
-            "version":1,
-            "timeline_layers":["000000000000000000000000000000000000-FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF__0000000001696070-00000000016960E9"],
-            "missing_layers":["not_a_real_layer_but_adding_coverage"],
-            "layer_metadata":{
-                "000000000000000000000000000000000000-FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF__0000000001696070-00000000016960E9": { "file_size": 25600000 },
-                "not_a_real_layer_but_adding_coverage": { "file_size": 9007199254741001 }
-            },
-            "disk_consistent_lsn":"0/16960E8",
-            "metadata_bytes":[113,11,159,210,0,54,0,4,0,0,0,0,1,105,96,232,1,0,0,0,0,1,105,96,112,0,0,0,0,0,0,0,0,0,0,0,0,0,1,105,96,112,0,0,0,0,1,105,96,112,0,0,0,14,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0]
-        }"#;
-
-        let expected = IndexPart {
-            // note this is not verified, could be anything, but exists for humans debugging.. could be the git version instead?
-            version: 1,
-            timeline_layers: [RelativePath("000000000000000000000000000000000000-FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF__0000000001696070-00000000016960E9".to_owned())].into_iter().collect(),
-            missing_layers: [RelativePath("not_a_real_layer_but_adding_coverage".to_owned())].into_iter().collect(),
-            layer_metadata: HashMap::from([
-                (RelativePath("000000000000000000000000000000000000-FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF__0000000001696070-00000000016960E9".to_owned()), IndexLayerMetadata {
-                    file_size: Some(25600000),
-                }),
-                (RelativePath("not_a_real_layer_but_adding_coverage".to_owned()), IndexLayerMetadata {
-                    // serde_json should always parse this but this might be a double with jq for
-                    // example.
-                    file_size: Some(9007199254741001),
-                })
-            ]),
-            disk_consistent_lsn: "0/16960E8".parse::<Lsn>().unwrap(),
-            metadata_bytes: [113,11,159,210,0,54,0,4,0,0,0,0,1,105,96,232,1,0,0,0,0,1,105,96,112,0,0,0,0,0,0,0,0,0,0,0,0,0,1,105,96,112,0,0,0,0,1,105,96,112,0,0,0,14,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0].to_vec(),
-        };
-
-        let part = serde_json::from_str::<IndexPart>(example).unwrap();
-        assert_eq!(part, expected);
-    }
 }

pageserver/src/storage_sync/upload.rs

@@ -69,25 +69,14 @@ pub(super) async fn upload_timeline_layers<'a>(
         .map(|meta| meta.disk_consistent_lsn());
 
     let already_uploaded_layers = remote_timeline
-        .map(|timeline| {
-            timeline
-                .stored_files()
-                .keys()
-                .cloned()
-                .collect::<std::collections::HashSet<_>>()
-        })
+        .map(|timeline| timeline.stored_files())
+        .cloned()
         .unwrap_or_default();
 
     let layers_to_upload = upload
         .layers_to_upload
-        .iter()
-        .filter_map(|(k, v)| {
-            if !already_uploaded_layers.contains(k) {
-                Some((k.to_owned(), v.to_owned()))
-            } else {
-                None
-            }
-        })
+        .difference(&already_uploaded_layers)
+        .cloned()
         .collect::<Vec<_>>();
 
     if layers_to_upload.is_empty() {
@@ -109,7 +98,7 @@ pub(super) async fn upload_timeline_layers<'a>(
 
     let mut upload_tasks = layers_to_upload
         .into_iter()
-        .map(|(source_path, known_metadata)| async move {
+        .map(|source_path| async move {
             let source_file = match fs::File::open(&source_path).await.with_context(|| {
                 format!(
                     "Failed to upen a source file for layer '{}'",
@@ -120,7 +109,7 @@ pub(super) async fn upload_timeline_layers<'a>(
                 Err(e) => return Err(UploadError::MissingLocalFile(source_path, e)),
             };
 
-            let fs_size = source_file
+            let source_size = source_file
                 .metadata()
                 .await
                 .with_context(|| {
@@ -130,24 +119,10 @@ pub(super) async fn upload_timeline_layers<'a>(
                     )
                 })
                 .map_err(UploadError::Other)?
-                .len();
-
-            // FIXME: this looks bad
-            if let Some(metadata_size) = known_metadata.file_size() {
-                if metadata_size != fs_size {
-                    return Err(UploadError::Other(anyhow::anyhow!(
-                        "File {source_path:?} has its current FS size {fs_size} diferent from initially determined {metadata_size}"
-                    )));
-                }
-            } else {
-                // this is a silly state we would like to avoid
-            }
-
-            let fs_size = usize::try_from(fs_size).with_context(|| format!("File {source_path:?} size {fs_size} could not be converted to usize"))
-                .map_err(UploadError::Other)?;
+                .len() as usize;
 
             match storage
-                .upload_storage_object(Box::new(source_file), fs_size, &source_path)
+                .upload_storage_object(Box::new(source_file), source_size, &source_path)
                 .await
                 .with_context(|| format!("Failed to upload layer file for {sync_id}"))
             {
@@ -161,11 +136,8 @@ pub(super) async fn upload_timeline_layers<'a>(
     while let Some(upload_result) = upload_tasks.next().await {
         match upload_result {
             Ok(uploaded_path) => {
-                let metadata = upload
-                    .layers_to_upload
-                    .remove(&uploaded_path)
-                    .expect("metadata should always exist, assuming no double uploads");
-                upload.uploaded_layers.insert(uploaded_path, metadata);
+                upload.layers_to_upload.remove(&uploaded_path);
+                upload.uploaded_layers.insert(uploaded_path);
             }
             Err(e) => match e {
                 UploadError::Other(e) => {
@@ -290,7 +262,7 @@ mod tests {
         assert_eq!(
             upload
                 .uploaded_layers
-                .keys()
+                .iter()
                 .cloned()
                 .collect::<BTreeSet<_>>(),
             layer_files
@@ -385,7 +357,7 @@ mod tests {
         assert_eq!(
             upload
                 .uploaded_layers
-                .keys()
+                .iter()
                 .cloned()
                 .collect::<BTreeSet<_>>(),
             layer_files

pageserver/src/tenant/layer_map.rs

@@ -62,8 +62,6 @@ pub struct LayerMap {
 
 struct LayerRTreeObject {
     layer: Arc<dyn Layer>,
-
-    envelope: AABB<[IntKey; 2]>,
 }
 
 // Representation of Key as numeric type.
@@ -199,16 +197,9 @@ impl PartialEq for LayerRTreeObject {
 impl RTreeObject for LayerRTreeObject {
     type Envelope = AABB<[IntKey; 2]>;
     fn envelope(&self) -> Self::Envelope {
-        self.envelope
-    }
-}
-
-impl LayerRTreeObject {
-    fn new(layer: Arc<dyn Layer>) -> Self {
-        let key_range = layer.get_key_range();
-        let lsn_range = layer.get_lsn_range();
-
-        let envelope = AABB::from_corners(
+        let key_range = self.layer.get_key_range();
+        let lsn_range = self.layer.get_lsn_range();
+        AABB::from_corners(
             [
                 IntKey::from(key_range.start.to_i128()),
                 IntKey::from(lsn_range.start.0 as i128),
@@ -217,8 +208,7 @@ impl LayerRTreeObject {
                 IntKey::from(key_range.end.to_i128() - 1),
                 IntKey::from(lsn_range.end.0 as i128 - 1),
             ], // AABB::upper is inclusive, while `key_range.end` and `lsn_range.end` are exclusive
-        );
-        LayerRTreeObject { layer, envelope }
+        )
     }
 }
 
@@ -348,7 +338,7 @@ impl LayerMap {
         if layer.get_key_range() == (Key::MIN..Key::MAX) {
             self.l0_delta_layers.push(layer.clone());
         }
-        self.historic_layers.insert(LayerRTreeObject::new(layer));
+        self.historic_layers.insert(LayerRTreeObject { layer });
         NUM_ONDISK_LAYERS.inc();
     }
 
@@ -372,7 +362,7 @@ impl LayerMap {
         }
         assert!(self
             .historic_layers
-            .remove(&LayerRTreeObject::new(layer))
+            .remove(&LayerRTreeObject { layer })
             .is_some());
         NUM_ONDISK_LAYERS.dec();
     }

pageserver/src/tenant/timeline.rs

@@ -1,12 +1,10 @@
 //!
 
-use anyhow::{anyhow, bail, ensure, Context};
+use anyhow::{anyhow, bail, ensure, Context, Result};
 use bytes::Bytes;
 use fail::fail_point;
 use itertools::Itertools;
 use once_cell::sync::OnceCell;
-use pageserver_api::models::TimelineState;
-use tokio::sync::watch;
 use tokio::task::spawn_blocking;
 use tracing::*;
 
@@ -34,12 +32,10 @@ use crate::tenant::{
 use crate::config::{PageServerConf, METADATA_FILE_NAME};
 use crate::keyspace::{KeyPartitioning, KeySpace};
 use crate::metrics::TimelineMetrics;
-use crate::page_image_cache;
 use crate::pgdatadir_mapping::BlockNumber;
 use crate::pgdatadir_mapping::LsnForTimestamp;
-use crate::pgdatadir_mapping::{is_rel_fsm_block_key, is_rel_vm_block_key};
+use crate::reltag::RelTag;
 use crate::tenant_config::TenantConfOpt;
-use pageserver_api::reltag::RelTag;
 
 use postgres_ffi::to_pg_timestamp;
 use utils::{
@@ -56,11 +52,7 @@ use crate::task_mgr::TaskKind;
 use crate::walreceiver::{is_etcd_client_initialized, spawn_connection_manager_task};
 use crate::walredo::WalRedoManager;
 use crate::CheckpointConfig;
-use crate::ZERO_PAGE;
-use crate::{
-    page_cache,
-    storage_sync::{self, index::LayerFileMetadata},
-};
+use crate::{page_cache, storage_sync};
 
 pub struct Timeline {
     conf: &'static PageServerConf,
@@ -163,8 +155,6 @@ pub struct Timeline {
 
     /// Relation size cache
     pub rel_size_cache: RwLock<HashMap<RelTag, (Lsn, BlockNumber)>>,
-
-    state: watch::Sender<TimelineState>,
 }
 
 /// Internal structure to hold all data needed for logical size calculation.
@@ -312,6 +302,10 @@ pub struct GcInfo {
 
 /// Public interface functions
 impl Timeline {
+    //------------------------------------------------------------------------------
+    // Public GET functions
+    //------------------------------------------------------------------------------
+
     /// Get the LSN where this branch was created
     pub fn get_ancestor_lsn(&self) -> Lsn {
         self.ancestor_lsn
@@ -421,11 +415,9 @@ impl Timeline {
     /// those functions with an LSN that has been processed yet is an error.
     ///
     pub async fn wait_lsn(&self, lsn: Lsn) -> anyhow::Result<()> {
-        anyhow::ensure!(self.is_active(), "Cannot wait for Lsn on inactive timeline");
-
         // This should never be called from the WAL receiver, because that could lead
         // to a deadlock.
-        anyhow::ensure!(
+        ensure!(
             task_mgr::current_task_kind() != Some(TaskKind::WalReceiverConnection),
             "wait_lsn cannot be called in WAL receiver"
         );
@@ -448,7 +440,7 @@ impl Timeline {
         &self,
         lsn: Lsn,
         latest_gc_cutoff_lsn: &RcuReadGuard<Lsn>,
-    ) -> anyhow::Result<()> {
+    ) -> Result<()> {
         ensure!(
             lsn >= **latest_gc_cutoff_lsn,
             "LSN {} is earlier than latest GC horizon {} (we might've already garbage collected needed data)",
@@ -458,6 +450,12 @@ impl Timeline {
         Ok(())
     }
 
+    //------------------------------------------------------------------------------
+    // Public PUT functions, to update the repository with new page versions.
+    //
+    // These are called by the WAL receiver to digest WAL records.
+    //------------------------------------------------------------------------------
+
     /// Flush to disk all data that was written with the put_* functions
     ///
     /// NOTE: This has nothing to do with checkpoint in PostgreSQL. We don't
@@ -476,201 +474,17 @@ impl Timeline {
         }
     }
 
-    pub fn compact(&self) -> anyhow::Result<()> {
-        let last_record_lsn = self.get_last_record_lsn();
-
-        // Last record Lsn could be zero in case the timelie was just created
-        if !last_record_lsn.is_valid() {
-            warn!("Skipping compaction for potentially just initialized timeline, it has invalid last record lsn: {last_record_lsn}");
-            return Ok(());
-        }
-
-        //
-        // High level strategy for compaction / image creation:
-        //
-        // 1. First, calculate the desired "partitioning" of the
-        // currently in-use key space. The goal is to partition the
-        // key space into roughly fixed-size chunks, but also take into
-        // account any existing image layers, and try to align the
-        // chunk boundaries with the existing image layers to avoid
-        // too much churn. Also try to align chunk boundaries with
-        // relation boundaries.  In principle, we don't know about
-        // relation boundaries here, we just deal with key-value
-        // pairs, and the code in pgdatadir_mapping.rs knows how to
-        // map relations into key-value pairs. But in practice we know
-        // that 'field6' is the block number, and the fields 1-5
-        // identify a relation. This is just an optimization,
-        // though.
-        //
-        // 2. Once we know the partitioning, for each partition,
-        // decide if it's time to create a new image layer. The
-        // criteria is: there has been too much "churn" since the last
-        // image layer? The "churn" is fuzzy concept, it's a
-        // combination of too many delta files, or too much WAL in
-        // total in the delta file. Or perhaps: if creating an image
-        // file would allow to delete some older files.
-        //
-        // 3. After that, we compact all level0 delta files if there
-        // are too many of them.  While compacting, we also garbage
-        // collect any page versions that are no longer needed because
-        // of the new image layers we created in step 2.
-        //
-        // TODO: This high level strategy hasn't been implemented yet.
-        // Below are functions compact_level0() and create_image_layers()
-        // but they are a bit ad hoc and don't quite work like it's explained
-        // above. Rewrite it.
-        let _layer_removal_cs = self.layer_removal_cs.lock().unwrap();
-
-        let target_file_size = self.get_checkpoint_distance();
-
-        // Define partitioning schema if needed
-
-        match self.repartition(
-            self.get_last_record_lsn(),
-            self.get_compaction_target_size(),
-        ) {
-            Ok((partitioning, lsn)) => {
-                // 2. Create new image layers for partitions that have been modified
-                // "enough".
-                let layer_paths_to_upload = self.create_image_layers(&partitioning, lsn, false)?;
-                if !layer_paths_to_upload.is_empty()
-                    && self.upload_layers.load(atomic::Ordering::Relaxed)
-                {
-                    storage_sync::schedule_layer_upload(
-                        self.tenant_id,
-                        self.timeline_id,
-                        layer_paths_to_upload,
-                        None,
-                    );
-                }
-
-                // 3. Compact
-                let timer = self.metrics.compact_time_histo.start_timer();
-                self.compact_level0(target_file_size)?;
-                timer.stop_and_record();
-            }
-            Err(err) => {
-                // no partitioning? This is normal, if the timeline was just created
-                // as an empty timeline. Also in unit tests, when we use the timeline
-                // as a simple key-value store, ignoring the datadir layout. Log the
-                // error but continue.
-                error!("could not compact, repartitioning keyspace failed: {err:?}");
-            }
-        };
-
-        Ok(())
-    }
-
     /// Mutate the timeline with a [`TimelineWriter`].
+    ///
+    /// FIXME: This ought to return &'a TimelineWriter, where TimelineWriter
+    /// is a generic type in this trait. But that doesn't currently work in
+    /// Rust: https://rust-lang.github.io/rfcs/1598-generic_associated_types.html
     pub fn writer(&self) -> TimelineWriter<'_> {
         TimelineWriter {
             tl: self,
             _write_guard: self.write_lock.lock().unwrap(),
         }
     }
-
-    /// Retrieve current logical size of the timeline.
-    ///
-    /// The size could be lagging behind the actual number, in case
-    /// the initial size calculation has not been run (gets triggered on the first size access).
-    pub fn get_current_logical_size(self: &Arc<Self>) -> anyhow::Result<u64> {
-        let current_size = self.current_logical_size.current_size()?;
-        debug!("Current size: {current_size:?}");
-
-        let size = current_size.size();
-        if let (CurrentLogicalSize::Approximate(_), Some(init_lsn)) =
-            (current_size, self.current_logical_size.initial_part_end)
-        {
-            self.try_spawn_size_init_task(init_lsn);
-        }
-
-        Ok(size)
-    }
-
-    /// Check if more than 'checkpoint_distance' of WAL has been accumulated in
-    /// the in-memory layer, and initiate flushing it if so.
-    ///
-    /// Also flush after a period of time without new data -- it helps
-    /// safekeepers to regard pageserver as caught up and suspend activity.
-    pub fn check_checkpoint_distance(self: &Arc<Timeline>) -> anyhow::Result<()> {
-        let last_lsn = self.get_last_record_lsn();
-        let layers = self.layers.read().unwrap();
-        if let Some(open_layer) = &layers.open_layer {
-            let open_layer_size = open_layer.size()?;
-            drop(layers);
-            let last_freeze_at = self.last_freeze_at.load();
-            let last_freeze_ts = *(self.last_freeze_ts.read().unwrap());
-            let distance = last_lsn.widening_sub(last_freeze_at);
-            // Checkpointing the open layer can be triggered by layer size or LSN range.
-            // S3 has a 5 GB limit on the size of one upload (without multi-part upload), and
-            // we want to stay below that with a big margin.  The LSN distance determines how
-            // much WAL the safekeepers need to store.
-            if distance >= self.get_checkpoint_distance().into()
-                || open_layer_size > self.get_checkpoint_distance()
-                || (distance > 0 && last_freeze_ts.elapsed() >= self.get_checkpoint_timeout())
-            {
-                info!(
-                    "check_checkpoint_distance {}, layer size {}, elapsed since last flush {:?}",
-                    distance,
-                    open_layer_size,
-                    last_freeze_ts.elapsed()
-                );
-
-                self.freeze_inmem_layer(true);
-                self.last_freeze_at.store(last_lsn);
-                *(self.last_freeze_ts.write().unwrap()) = Instant::now();
-
-                // Launch a task to flush the frozen layer to disk, unless
-                // a task was already running. (If the task was running
-                // at the time that we froze the layer, it must've seen the
-                // the layer we just froze before it exited; see comments
-                // in flush_frozen_layers())
-                if let Ok(guard) = self.layer_flush_lock.try_lock() {
-                    drop(guard);
-                    let self_clone = Arc::clone(self);
-                    task_mgr::spawn(
-                        task_mgr::BACKGROUND_RUNTIME.handle(),
-                        task_mgr::TaskKind::LayerFlushTask,
-                        Some(self.tenant_id),
-                        Some(self.timeline_id),
-                        "layer flush task",
-                        false,
-                        async move { self_clone.flush_frozen_layers(false) },
-                    );
-                }
-            }
-        }
-        Ok(())
-    }
-
-    pub fn set_state(&self, new_state: TimelineState) {
-        match (self.current_state(), new_state) {
-            (equal_state_1, equal_state_2) if equal_state_1 == equal_state_2 => {
-                debug!("Ignoring new state, equal to the existing one: {equal_state_2:?}");
-            }
-            (TimelineState::Broken, _) => {
-                error!("Ignoring state update {new_state:?} for broken tenant");
-            }
-            (TimelineState::Paused, TimelineState::Active) => {
-                debug!("Not activating a paused timeline");
-            }
-            (_, new_state) => {
-                self.state.send_replace(new_state);
-            }
-        }
-    }
-
-    pub fn current_state(&self) -> TimelineState {
-        *self.state.borrow()
-    }
-
-    pub fn is_active(&self) -> bool {
-        self.current_state() == TimelineState::Active
-    }
-
-    pub fn subscribe_for_state_updates(&self) -> watch::Receiver<TimelineState> {
-        self.state.subscribe()
-    }
 }
 
 // Private functions
@@ -714,7 +528,7 @@ impl Timeline {
     ///
     /// Loads the metadata for the timeline into memory, but not the layer map.
     #[allow(clippy::too_many_arguments)]
-    pub(super) fn new(
+    pub fn new(
         conf: &'static PageServerConf,
         tenant_conf: Arc<RwLock<TenantConfOpt>>,
         metadata: TimelineMetadata,
@@ -724,9 +538,8 @@ impl Timeline {
         walredo_mgr: Arc<dyn WalRedoManager + Send + Sync>,
         upload_layers: bool,
         pg_version: u32,
-    ) -> Self {
+    ) -> Timeline {
         let disk_consistent_lsn = metadata.disk_consistent_lsn();
-        let (state, _) = watch::channel(TimelineState::Suspended);
 
         let mut result = Timeline {
             conf,
@@ -783,17 +596,16 @@ impl Timeline {
 
             last_received_wal: Mutex::new(None),
             rel_size_cache: RwLock::new(HashMap::new()),
-            state,
         };
         result.repartition_threshold = result.get_checkpoint_distance() / 10;
         result
     }
 
-    pub(super) fn launch_wal_receiver(self: &Arc<Self>) {
+    pub fn launch_wal_receiver(self: &Arc<Self>) -> anyhow::Result<()> {
         if !is_etcd_client_initialized() {
             if cfg!(test) {
                 info!("not launching WAL receiver because etcd client hasn't been initialized");
-                return;
+                return Ok(());
             } else {
                 panic!("etcd client not initialized");
             }
@@ -821,14 +633,16 @@ impl Timeline {
             walreceiver_connect_timeout,
             lagging_wal_timeout,
             max_lsn_wal_lag,
-        );
+        )?;
+
+        Ok(())
     }
 
     ///
     /// Scan the timeline directory to populate the layer map.
     /// Returns all timeline-related files that were found and loaded.
     ///
-    pub(super) fn load_layer_map(&self, disk_consistent_lsn: Lsn) -> anyhow::Result<()> {
+    pub fn load_layer_map(&self, disk_consistent_lsn: Lsn) -> anyhow::Result<()> {
         let mut layers = self.layers.write().unwrap();
         let mut num_layers = 0;
 
@@ -914,13 +728,33 @@ impl Timeline {
         Ok(())
     }
 
-    pub(super) fn layer_removal_guard(&self) -> anyhow::Result<MutexGuard<()>> {
+    pub fn layer_removal_guard(&self) -> anyhow::Result<MutexGuard<()>> {
         self.layer_removal_cs
             .try_lock()
             .map_err(|e| anyhow!("cannot lock compaction critical section {e}"))
     }
 
+    /// Retrieve current logical size of the timeline.
+    ///
+    /// The size could be lagging behind the actual number, in case
+    /// the initial size calculation has not been run (gets triggered on the first size access).
+    pub fn get_current_logical_size(self: &Arc<Self>) -> anyhow::Result<u64> {
+        let current_size = self.current_logical_size.current_size()?;
+        debug!("Current size: {current_size:?}");
+
+        let size = current_size.size();
+        if let (CurrentLogicalSize::Approximate(_), Some(init_lsn)) =
+            (current_size, self.current_logical_size.initial_part_end)
+        {
+            self.try_spawn_size_init_task(init_lsn);
+        }
+
+        Ok(size)
+    }
+
     fn try_spawn_size_init_task(self: &Arc<Self>, init_lsn: Lsn) {
+        let timeline_id = self.timeline_id;
+
         // Atomically check if the timeline size calculation had already started.
         // If the flag was not already set, this sets it.
         if !self
@@ -937,42 +771,17 @@ impl Timeline {
                 "initial size calculation",
                 false,
                 async move {
-                    let mut timeline_state_updates = self_clone.subscribe_for_state_updates();
-                    let self_calculation = Arc::clone(&self_clone);
-                    tokio::select! {
-                        calculation_result = spawn_blocking(move || self_calculation.calculate_logical_size(init_lsn)) => {
-                            let calculated_size = calculation_result
-                                .context("Failed to spawn calculation result task")?
-                                .context("Failed to calculate logical size")?;
-                            match self_clone.current_logical_size.initial_logical_size.set(calculated_size) {
-                                Ok(()) => info!("Successfully calculated initial logical size"),
-                                Err(existing_size) => error!("Tried to update initial timeline size value to {calculated_size}, but the size was already set to {existing_size}, not changing"),
-                            }
-                            Ok(())
-                        },
-                        new_event = async {
-                            loop {
-                                match timeline_state_updates.changed().await {
-                                    Ok(()) => {
-                                        let new_state = *timeline_state_updates.borrow();
-                                        match new_state {
-                                            // we're running this job for active timelines only
-                                            TimelineState::Active => continue,
-                                            TimelineState::Broken | TimelineState::Paused | TimelineState::Suspended => return Some(new_state),
-                                        }
-                                    }
-                                    Err(_sender_dropped_error) => return None,
-                                }
-                            }
-                        } => {
-                            match new_event {
-                                Some(new_state) => info!("Timeline became inactive (new state: {new_state:?}), dropping current connections until it reactivates"),
-                                None => info!("Timeline dropped state updates sender, stopping init size calculation"),
-                            }
-                            Ok(())
-                        },
+                    let calculated_size = self_clone.calculate_logical_size(init_lsn)?;
+                    let result = spawn_blocking(move || {
+                        self_clone.current_logical_size.initial_logical_size.set(calculated_size)
+                    }).await?;
+                    match result {
+                        Ok(()) => info!("Successfully calculated initial logical size"),
+                        Err(existing_size) => error!("Tried to update initial timeline size value to {calculated_size}, but the size was already set to {existing_size}, not changing"),
                     }
-                }.instrument(info_span!("initial_logical_size_calculation", tenant = %self.tenant_id, timeline = %self.timeline_id)),
+                    Ok(())
+                }
+                .instrument(info_span!("initial_logical_size_calculation", timeline = %timeline_id))
             );
         }
     }
@@ -1163,7 +972,7 @@ impl Timeline {
         Some((lsn, img))
     }
 
-    fn get_ancestor_timeline(&self) -> anyhow::Result<Arc<Timeline>> {
+    fn get_ancestor_timeline(&self) -> Result<Arc<Timeline>> {
         let ancestor = self.ancestor_timeline.as_ref().with_context(|| {
             format!(
                 "Ancestor is missing. Timeline id: {} Ancestor id {:?}",
@@ -1222,14 +1031,14 @@ impl Timeline {
         Ok(layer)
     }
 
-    fn put_value(&self, key: Key, lsn: Lsn, val: &Value) -> anyhow::Result<()> {
+    fn put_value(&self, key: Key, lsn: Lsn, val: &Value) -> Result<()> {
         //info!("PUT: key {} at {}", key, lsn);
         let layer = self.get_layer_for_write(lsn)?;
         layer.put_value(key, lsn, val)?;
         Ok(())
     }
 
-    fn put_tombstone(&self, key_range: Range<Key>, lsn: Lsn) -> anyhow::Result<()> {
+    fn put_tombstone(&self, key_range: Range<Key>, lsn: Lsn) -> Result<()> {
         let layer = self.get_layer_for_write(lsn)?;
         layer.put_tombstone(key_range, lsn)?;
 
@@ -1268,6 +1077,64 @@ impl Timeline {
         drop(layers);
     }
 
+    ///
+    /// Check if more than 'checkpoint_distance' of WAL has been accumulated in
+    /// the in-memory layer, and initiate flushing it if so.
+    ///
+    /// Also flush after a period of time without new data -- it helps
+    /// safekeepers to regard pageserver as caught up and suspend activity.
+    ///
+    pub fn check_checkpoint_distance(self: &Arc<Timeline>) -> Result<()> {
+        let last_lsn = self.get_last_record_lsn();
+        let layers = self.layers.read().unwrap();
+        if let Some(open_layer) = &layers.open_layer {
+            let open_layer_size = open_layer.size()?;
+            drop(layers);
+            let last_freeze_at = self.last_freeze_at.load();
+            let last_freeze_ts = *(self.last_freeze_ts.read().unwrap());
+            let distance = last_lsn.widening_sub(last_freeze_at);
+            // Checkpointing the open layer can be triggered by layer size or LSN range.
+            // S3 has a 5 GB limit on the size of one upload (without multi-part upload), and
+            // we want to stay below that with a big margin.  The LSN distance determines how
+            // much WAL the safekeepers need to store.
+            if distance >= self.get_checkpoint_distance().into()
+                || open_layer_size > self.get_checkpoint_distance()
+                || (distance > 0 && last_freeze_ts.elapsed() >= self.get_checkpoint_timeout())
+            {
+                info!(
+                    "check_checkpoint_distance {}, layer size {}, elapsed since last flush {:?}",
+                    distance,
+                    open_layer_size,
+                    last_freeze_ts.elapsed()
+                );
+
+                self.freeze_inmem_layer(true);
+                self.last_freeze_at.store(last_lsn);
+                *(self.last_freeze_ts.write().unwrap()) = Instant::now();
+
+                // Launch a task to flush the frozen layer to disk, unless
+                // a task was already running. (If the task was running
+                // at the time that we froze the layer, it must've seen the
+                // the layer we just froze before it exited; see comments
+                // in flush_frozen_layers())
+                if let Ok(guard) = self.layer_flush_lock.try_lock() {
+                    drop(guard);
+                    let self_clone = Arc::clone(self);
+                    task_mgr::spawn(
+                        task_mgr::BACKGROUND_RUNTIME.handle(),
+                        task_mgr::TaskKind::LayerFlushTask,
+                        Some(self.tenant_id),
+                        Some(self.timeline_id),
+                        "layer flush task",
+                        false,
+                        async move { self_clone.flush_frozen_layers(false) },
+                    );
+                }
+            }
+        }
+        Ok(())
+    }
+
     /// Flush all frozen layers to disk.
     ///
     /// Only one task at a time can be doing layer-flushing for a
@@ -1275,7 +1142,7 @@ impl Timeline {
     /// currently doing the flushing, this function will wait for it
     /// to finish. If 'wait' is false, this function will return
     /// immediately instead.
-    fn flush_frozen_layers(&self, wait: bool) -> anyhow::Result<()> {
+    fn flush_frozen_layers(&self, wait: bool) -> Result<()> {
         let flush_lock_guard = if wait {
             self.layer_flush_lock.lock().unwrap()
         } else {
@@ -1314,7 +1181,7 @@ impl Timeline {
     }
 
     /// Flush one frozen in-memory layer to disk, as a new delta layer.
-    fn flush_frozen_layer(&self, frozen_layer: Arc<InMemoryLayer>) -> anyhow::Result<()> {
+    fn flush_frozen_layer(&self, frozen_layer: Arc<InMemoryLayer>) -> Result<()> {
         // As a special case, when we have just imported an image into the repository,
         // instead of writing out a L0 delta layer, we directly write out image layer
         // files instead. This is possible as long as *all* the data imported into the
@@ -1327,8 +1194,8 @@ impl Timeline {
                 self.create_image_layers(&partitioning, self.initdb_lsn, true)?
             } else {
                 // normal case, write out a L0 delta layer file.
-                let (delta_path, metadata) = self.create_delta_layer(&frozen_layer)?;
-                HashMap::from([(delta_path, metadata)])
+                let delta_path = self.create_delta_layer(&frozen_layer)?;
+                HashSet::from([delta_path])
             };
 
         fail_point!("flush-frozen-before-sync");
@@ -1354,86 +1221,85 @@ impl Timeline {
         // TODO: This perhaps should be done in 'flush_frozen_layers', after flushing
         // *all* the layers, to avoid fsyncing the file multiple times.
         let disk_consistent_lsn = Lsn(lsn_range.end.0 - 1);
-        let old_disk_consistent_lsn = self.disk_consistent_lsn.load();
+        self.update_disk_consistent_lsn(disk_consistent_lsn, layer_paths_to_upload)?;
 
-        // If we were able to advance 'disk_consistent_lsn', save it the metadata file.
-        // After crash, we will restart WAL streaming and processing from that point.
-        if disk_consistent_lsn != old_disk_consistent_lsn {
-            assert!(disk_consistent_lsn > old_disk_consistent_lsn);
-            self.update_metadata_file(disk_consistent_lsn, layer_paths_to_upload)?;
-            // Also update the in-memory copy
-            self.disk_consistent_lsn.store(disk_consistent_lsn);
-        }
         Ok(())
     }
 
     /// Update metadata file
-    fn update_metadata_file(
+    fn update_disk_consistent_lsn(
         &self,
         disk_consistent_lsn: Lsn,
-        layer_paths_to_upload: HashMap<PathBuf, LayerFileMetadata>,
-    ) -> anyhow::Result<()> {
-        // We can only save a valid 'prev_record_lsn' value on disk if we
-        // flushed *all* in-memory changes to disk. We only track
-        // 'prev_record_lsn' in memory for the latest processed record, so we
-        // don't remember what the correct value that corresponds to some old
-        // LSN is. But if we flush everything, then the value corresponding
-        // current 'last_record_lsn' is correct and we can store it on disk.
-        let RecordLsn {
-            last: last_record_lsn,
-            prev: prev_record_lsn,
-        } = self.last_record_lsn.load();
-        let ondisk_prev_record_lsn = if disk_consistent_lsn == last_record_lsn {
-            Some(prev_record_lsn)
-        } else {
-            None
-        };
+        layer_paths_to_upload: HashSet<PathBuf>,
+    ) -> Result<()> {
+        // If we were able to advance 'disk_consistent_lsn', save it the metadata file.
+        // After crash, we will restart WAL streaming and processing from that point.
+        let old_disk_consistent_lsn = self.disk_consistent_lsn.load();
+        if disk_consistent_lsn != old_disk_consistent_lsn {
+            assert!(disk_consistent_lsn > old_disk_consistent_lsn);
 
-        let ancestor_timeline_id = self
-            .ancestor_timeline
-            .as_ref()
-            .map(|ancestor| ancestor.timeline_id);
+            // We can only save a valid 'prev_record_lsn' value on disk if we
+            // flushed *all* in-memory changes to disk. We only track
+            // 'prev_record_lsn' in memory for the latest processed record, so we
+            // don't remember what the correct value that corresponds to some old
+            // LSN is. But if we flush everything, then the value corresponding
+            // current 'last_record_lsn' is correct and we can store it on disk.
+            let RecordLsn {
+                last: last_record_lsn,
+                prev: prev_record_lsn,
+            } = self.last_record_lsn.load();
+            let ondisk_prev_record_lsn = if disk_consistent_lsn == last_record_lsn {
+                Some(prev_record_lsn)
+            } else {
+                None
+            };
 
-        let metadata = TimelineMetadata::new(
-            disk_consistent_lsn,
-            ondisk_prev_record_lsn,
-            ancestor_timeline_id,
-            self.ancestor_lsn,
-            *self.latest_gc_cutoff_lsn.read(),
-            self.initdb_lsn,
-            self.pg_version,
-        );
+            let ancestor_timeline_id = self
+                .ancestor_timeline
+                .as_ref()
+                .map(|ancestor| ancestor.timeline_id);
 
-        fail_point!("checkpoint-before-saving-metadata", |x| bail!(
-            "{}",
-            x.unwrap()
-        ));
-
-        save_metadata(
-            self.conf,
-            self.timeline_id,
-            self.tenant_id,
-            &metadata,
-            false,
-        )?;
-
-        if self.can_upload_layers() {
-            storage_sync::schedule_layer_upload(
-                self.tenant_id,
-                self.timeline_id,
-                layer_paths_to_upload,
-                Some(metadata),
+            let metadata = TimelineMetadata::new(
+                disk_consistent_lsn,
+                ondisk_prev_record_lsn,
+                ancestor_timeline_id,
+                self.ancestor_lsn,
+                *self.latest_gc_cutoff_lsn.read(),
+                self.initdb_lsn,
+                self.pg_version,
             );
+
+            fail_point!("checkpoint-before-saving-metadata", |x| bail!(
+                "{}",
+                x.unwrap()
+            ));
+
+            save_metadata(
+                self.conf,
+                self.timeline_id,
+                self.tenant_id,
+                &metadata,
+                false,
+            )?;
+
+            if self.upload_layers.load(atomic::Ordering::Relaxed) {
+                storage_sync::schedule_layer_upload(
+                    self.tenant_id,
+                    self.timeline_id,
+                    layer_paths_to_upload,
+                    Some(metadata),
+                );
+            }
+
+            // Also update the in-memory copy
+            self.disk_consistent_lsn.store(disk_consistent_lsn);
         }
 
         Ok(())
     }
 
     // Write out the given frozen in-memory layer as a new L0 delta file
-    fn create_delta_layer(
-        &self,
-        frozen_layer: &InMemoryLayer,
-    ) -> anyhow::Result<(PathBuf, LayerFileMetadata)> {
+    fn create_delta_layer(&self, frozen_layer: &InMemoryLayer) -> Result<PathBuf> {
         // Write it out
         let new_delta = frozen_layer.write_to_disk()?;
         let new_delta_path = new_delta.path();
@@ -1459,16 +1325,100 @@ impl Timeline {
 
         // update the timeline's physical size
         let sz = new_delta_path.metadata()?.len();
-
         self.metrics.current_physical_size_gauge.add(sz);
         // update metrics
         self.metrics.num_persistent_files_created.inc_by(1);
         self.metrics.persistent_bytes_written.inc_by(sz);
 
-        Ok((new_delta_path, LayerFileMetadata::new(sz)))
+        Ok(new_delta_path)
     }
 
-    fn repartition(&self, lsn: Lsn, partition_size: u64) -> anyhow::Result<(KeyPartitioning, Lsn)> {
+    pub fn compact(&self) -> anyhow::Result<()> {
+        let last_record_lsn = self.get_last_record_lsn();
+
+        // Last record Lsn could be zero in case the timelie was just created
+        if !last_record_lsn.is_valid() {
+            warn!("Skipping compaction for potentially just initialized timeline, it has invalid last record lsn: {last_record_lsn}");
+            return Ok(());
+        }
+
+        //
+        // High level strategy for compaction / image creation:
+        //
+        // 1. First, calculate the desired "partitioning" of the
+        // currently in-use key space. The goal is to partition the
+        // key space into roughly fixed-size chunks, but also take into
+        // account any existing image layers, and try to align the
+        // chunk boundaries with the existing image layers to avoid
+        // too much churn. Also try to align chunk boundaries with
+        // relation boundaries.  In principle, we don't know about
+        // relation boundaries here, we just deal with key-value
+        // pairs, and the code in pgdatadir_mapping.rs knows how to
+        // map relations into key-value pairs. But in practice we know
+        // that 'field6' is the block number, and the fields 1-5
+        // identify a relation. This is just an optimization,
+        // though.
+        //
+        // 2. Once we know the partitioning, for each partition,
+        // decide if it's time to create a new image layer. The
+        // criteria is: there has been too much "churn" since the last
+        // image layer? The "churn" is fuzzy concept, it's a
+        // combination of too many delta files, or too much WAL in
+        // total in the delta file. Or perhaps: if creating an image
+        // file would allow to delete some older files.
+        //
+        // 3. After that, we compact all level0 delta files if there
+        // are too many of them.  While compacting, we also garbage
+        // collect any page versions that are no longer needed because
+        // of the new image layers we created in step 2.
+        //
+        // TODO: This high level strategy hasn't been implemented yet.
+        // Below are functions compact_level0() and create_image_layers()
+        // but they are a bit ad hoc and don't quite work like it's explained
+        // above. Rewrite it.
+        let _layer_removal_cs = self.layer_removal_cs.lock().unwrap();
+
+        let target_file_size = self.get_checkpoint_distance();
+
+        // Define partitioning schema if needed
+
+        match self.repartition(
+            self.get_last_record_lsn(),
+            self.get_compaction_target_size(),
+        ) {
+            Ok((partitioning, lsn)) => {
+                // 2. Create new image layers for partitions that have been modified
+                // "enough".
+                let layer_paths_to_upload = self.create_image_layers(&partitioning, lsn, false)?;
+                if !layer_paths_to_upload.is_empty()
+                    && self.upload_layers.load(atomic::Ordering::Relaxed)
+                {
+                    storage_sync::schedule_layer_upload(
+                        self.tenant_id,
+                        self.timeline_id,
+                        HashSet::from_iter(layer_paths_to_upload),
+                        None,
+                    );
+                }
+
+                // 3. Compact
+                let timer = self.metrics.compact_time_histo.start_timer();
+                self.compact_level0(target_file_size)?;
+                timer.stop_and_record();
+            }
+            Err(err) => {
+                // no partitioning? This is normal, if the timeline was just created
+                // as an empty timeline. Also in unit tests, when we use the timeline
+                // as a simple key-value store, ignoring the datadir layout. Log the
+                // error but continue.
+                error!("could not compact, repartitioning keyspace failed: {err:?}");
+            }
+        };
+
+        Ok(())
+    }
+
+    fn repartition(&self, lsn: Lsn, partition_size: u64) -> Result<(KeyPartitioning, Lsn)> {
         let mut partitioning_guard = self.partitioning.lock().unwrap();
         if partitioning_guard.1 == Lsn(0)
             || lsn.0 - partitioning_guard.1 .0 > self.repartition_threshold
@@ -1482,7 +1432,7 @@ impl Timeline {
     }
 
     // Is it time to create a new image layer for the given partition?
-    fn time_for_new_image_layer(&self, partition: &KeySpace, lsn: Lsn) -> anyhow::Result<bool> {
+    fn time_for_new_image_layer(&self, partition: &KeySpace, lsn: Lsn) -> Result<bool> {
         let layers = self.layers.read().unwrap();
 
         for part_range in &partition.ranges {
@@ -1527,9 +1477,10 @@ impl Timeline {
         partitioning: &KeyPartitioning,
         lsn: Lsn,
         force: bool,
-    ) -> anyhow::Result<HashMap<PathBuf, LayerFileMetadata>> {
+    ) -> Result<HashSet<PathBuf>> {
         let timer = self.metrics.create_images_time_histo.start_timer();
         let mut image_layers: Vec<ImageLayer> = Vec::new();
+        let mut layer_paths_to_upload = HashSet::new();
         for partition in partitioning.parts.iter() {
             if force || self.time_for_new_image_layer(partition, lsn)? {
                 let img_range =
@@ -1545,37 +1496,13 @@ impl Timeline {
                 for range in &partition.ranges {
                     let mut key = range.start;
                     while key < range.end {
-                        let img = match self.get(key, lsn) {
-                            Ok(img) => img,
-                            Err(err) => {
-                                // If we fail to reconstruct a VM or FSM page, we can zero the
-                                // page without losing any actual user data. That seems better
-                                // than failing repeatedly and getting stuck.
-                                //
-                                // We had a bug at one point, where we truncated the FSM and VM
-                                // in the pageserver, but the Postgres didn't know about that
-                                // and continued to generate incremental WAL records for pages
-                                // that didn't exist in the pageserver. Trying to replay those
-                                // WAL records failed to find the previous image of the page.
-                                // This special case allows us to recover from that situation.
-                                // See https://github.com/neondatabase/neon/issues/2601.
-                                //
-                                // Unfortunately we cannot do this for the main fork, or for
-                                // any metadata keys, keys, as that would lead to actual data
-                                // loss.
-                                if is_rel_fsm_block_key(key) || is_rel_vm_block_key(key) {
-                                    warn!("could not reconstruct FSM or VM key {key}, filling with zeros: {err:?}");
-                                    ZERO_PAGE.clone()
-                                } else {
-                                    return Err(err);
-                                }
-                            }
-                        };
+                        let img = self.get(key, lsn)?;
                         image_layer_writer.put_image(key, &img)?;
                         key = key.next();
                     }
                 }
                 let image_layer = image_layer_writer.finish()?;
+                layer_paths_to_upload.insert(image_layer.path());
                 image_layers.push(image_layer);
             }
         }
@@ -1589,25 +1516,15 @@ impl Timeline {
         //
         // Compaction creates multiple image layers. It would be better to create them all
         // and fsync them all in parallel.
-        let all_paths = image_layers
-            .iter()
-            .map(|layer| layer.path())
-            .chain(std::iter::once(
-                self.conf.timeline_path(&self.timeline_id, &self.tenant_id),
-            ))
-            .collect::<Vec<_>>();
+        let mut all_paths = Vec::from_iter(layer_paths_to_upload.clone());
+        all_paths.push(self.conf.timeline_path(&self.timeline_id, &self.tenant_id));
         par_fsync::par_fsync(&all_paths)?;
 
-        let mut layer_paths_to_upload = HashMap::with_capacity(image_layers.len());
-
         let mut layers = self.layers.write().unwrap();
         for l in image_layers {
-            let path = l.path();
-            let metadata = path.metadata()?;
-
-            layer_paths_to_upload.insert(path, LayerFileMetadata::new(metadata.len()));
-
-            self.metrics.current_physical_size_gauge.add(metadata.len());
+            self.metrics
+                .current_physical_size_gauge
+                .add(l.path().metadata()?.len());
             layers.insert_historic(Arc::new(l));
         }
         drop(layers);
@@ -1620,7 +1537,7 @@ impl Timeline {
     /// Collect a bunch of Level 0 layer files, and compact and reshuffle them as
     /// as Level 1 files.
     ///
-    fn compact_level0(&self, target_file_size: u64) -> anyhow::Result<()> {
+    fn compact_level0(&self, target_file_size: u64) -> Result<()> {
         let layers = self.layers.read().unwrap();
         let mut level0_deltas = layers.get_level0_deltas()?;
         drop(layers);
@@ -1858,16 +1775,16 @@ impl Timeline {
         }
 
         let mut layers = self.layers.write().unwrap();
-        let mut new_layer_paths = HashMap::with_capacity(new_layers.len());
+        let mut new_layer_paths = HashSet::with_capacity(new_layers.len());
         for l in new_layers {
             let new_delta_path = l.path();
 
-            let metadata = new_delta_path.metadata()?;
-
             // update the timeline's physical size
-            self.metrics.current_physical_size_gauge.add(metadata.len());
+            self.metrics
+                .current_physical_size_gauge
+                .add(new_delta_path.metadata()?.len());
 
-            new_layer_paths.insert(new_delta_path, LayerFileMetadata::new(metadata.len()));
+            new_layer_paths.insert(new_delta_path);
             layers.insert_historic(Arc::new(l));
         }
 
@@ -1887,7 +1804,7 @@ impl Timeline {
         }
         drop(layers);
 
-        if self.can_upload_layers() {
+        if self.upload_layers.load(atomic::Ordering::Relaxed) {
             storage_sync::schedule_layer_upload(
                 self.tenant_id,
                 self.timeline_id,
@@ -1930,12 +1847,12 @@ impl Timeline {
     ///
     /// The 'pitr' duration is used to calculate a 'pitr_cutoff', which can be used to determine
     /// whether a record is needed for PITR.
-    pub(super) fn update_gc_info(
+    pub fn update_gc_info(
         &self,
         retain_lsns: Vec<Lsn>,
         cutoff_horizon: Lsn,
         pitr: Duration,
-    ) -> anyhow::Result<()> {
+    ) -> Result<()> {
         let mut gc_info = self.gc_info.write().unwrap();
 
         gc_info.horizon_cutoff = cutoff_horizon;
@@ -1990,8 +1907,8 @@ impl Timeline {
     /// within a layer file. We can only remove the whole file if it's fully
     /// obsolete.
     ///
-    pub(super) fn gc(&self) -> anyhow::Result<GcResult> {
-        let mut result: GcResult = GcResult::default();
+    pub fn gc(&self) -> Result<GcResult> {
+        let mut result: GcResult = Default::default();
         let now = SystemTime::now();
 
         fail_point!("before-timeline-gc");
@@ -2034,9 +1951,6 @@ impl Timeline {
             );
             write_guard.store_and_unlock(new_gc_cutoff).wait();
         }
-        // Persist the new GC cutoff value in the metadata file, before
-        // we actually remove anything.
-        self.update_metadata_file(self.disk_consistent_lsn.load(), HashMap::new())?;
 
         info!("GC starting");
 
@@ -2162,16 +2076,7 @@ impl Timeline {
             result.layers_removed += 1;
         }
 
-        info!(
-            "GC completed removing {} layers, cutoff {}",
-            result.layers_removed, new_gc_cutoff
-        );
-
-        if result.layers_removed != 0 {
-            fail_point!("after-timeline-gc-removed-layers");
-        }
-
-        if self.can_upload_layers() {
+        if self.upload_layers.load(atomic::Ordering::Relaxed) {
             storage_sync::schedule_layer_delete(
                 self.tenant_id,
                 self.timeline_id,
@@ -2260,11 +2165,6 @@ impl Timeline {
             }
         }
     }
-
-    fn can_upload_layers(&self) -> bool {
-        self.upload_layers.load(atomic::Ordering::Relaxed)
-            && self.current_state() != TimelineState::Broken
-    }
 }
 
 /// Helper function for get_reconstruct_data() to add the path of layers traversed
@@ -2315,12 +2215,11 @@ impl<'a> TimelineWriter<'a> {
     ///
     /// This will implicitly extend the relation, if the page is beyond the
     /// current end-of-file.
-    pub fn put(&self, key: Key, lsn: Lsn, value: &Value) -> anyhow::Result<()> {
-        page_image_cache::remove(key, self.tenant_id, self.timeline_id);
+    pub fn put(&self, key: Key, lsn: Lsn, value: &Value) -> Result<()> {
         self.tl.put_value(key, lsn, value)
     }
 
-    pub fn delete(&self, key_range: Range<Key>, lsn: Lsn) -> anyhow::Result<()> {
+    pub fn delete(&self, key_range: Range<Key>, lsn: Lsn) -> Result<()> {
         self.tl.put_tombstone(key_range, lsn)
     }
 

pageserver/src/tenant_mgr.rs

@@ -1,7 +1,7 @@
 //! This module acts as a switchboard to access different repositories managed by this
 //! page server.
 
-use std::collections::{hash_map, HashMap};
+use std::collections::{hash_map, HashMap, HashSet};
 use std::ffi::OsStr;
 use std::fs;
 use std::path::{Path, PathBuf};
@@ -12,19 +12,19 @@ use tracing::*;
 
 use remote_storage::GenericRemoteStorage;
 
-use crate::config::{PageServerConf, METADATA_FILE_NAME, TIMELINE_UNINIT_MARK_SUFFIX};
+use crate::config::{PageServerConf, METADATA_FILE_NAME};
 use crate::http::models::TenantInfo;
-use crate::storage_sync::index::{LayerFileMetadata, RemoteIndex, RemoteTimelineIndex};
-use crate::storage_sync::{self, LocalTimelineInitStatus, SyncStartupData, TimelineLocalFiles};
+use crate::storage_sync::index::{RemoteIndex, RemoteTimelineIndex};
+use crate::storage_sync::{self, LocalTimelineInitStatus, SyncStartupData};
 use crate::task_mgr::{self, TaskKind};
 use crate::tenant::{
     ephemeral_file::is_ephemeral_file, metadata::TimelineMetadata, Tenant, TenantState,
 };
 use crate::tenant_config::TenantConfOpt;
 use crate::walredo::PostgresRedoManager;
-use crate::TEMP_FILE_SUFFIX;
+use crate::{TenantTimelineValues, TEMP_FILE_SUFFIX};
 
-use utils::crashsafe::{self, path_with_suffix_extension};
+use utils::crashsafe_dir::{self, path_with_suffix_extension};
 use utils::id::{TenantId, TimelineId};
 
 mod tenants_state {
@@ -70,54 +70,34 @@ pub fn init_tenant_mgr(
             .remote_storage_config
             .as_ref()
             .expect("remote storage without config");
-        let mut broken_tenants = HashMap::new();
-        let mut ready_tenants = HashMap::new();
-        for (tenant_id, tenant_attach_data) in local_tenant_files.into_iter() {
-            match tenant_attach_data {
-                TenantAttachData::Ready(t) => {
-                    ready_tenants.insert(tenant_id, t);
-                }
-                TenantAttachData::Broken(e) => {
-                    broken_tenants.insert(tenant_id, TenantAttachData::Broken(e));
-                }
-            }
-        }
+
         let SyncStartupData {
             remote_index,
             local_timeline_init_statuses,
         } = storage_sync::spawn_storage_sync_task(
             conf,
-            ready_tenants,
+            local_tenant_files,
             storage,
             storage_config.max_concurrent_syncs,
             storage_config.max_sync_errors,
         )
         .context("Failed to spawn the storage sync thread")?;
 
-        let n = local_timeline_init_statuses.0.len();
-        let mut synced_timelines = local_timeline_init_statuses.0.into_iter().fold(
-            HashMap::<TenantId, TenantAttachData>::with_capacity(n),
-            |mut new_values, (tenant_id, old_values)| {
-                let new_timeline_values = new_values
-                    .entry(tenant_id)
-                    .or_insert_with(|| TenantAttachData::Ready(HashMap::new()));
-                if let TenantAttachData::Ready(t) = new_timeline_values {
-                    for (timeline_id, old_value) in old_values {
-                        if let LocalTimelineInitStatus::LocallyComplete(metadata) = old_value {
-                            t.insert(timeline_id, TimelineLocalFiles::ready(metadata));
-                        }
-                    }
-                }
-                new_values
-            },
-        );
-        synced_timelines.extend(broken_tenants);
-
-        (remote_index, synced_timelines)
+        (
+            remote_index,
+            local_timeline_init_statuses.filter_map(|init_status| match init_status {
+                LocalTimelineInitStatus::LocallyComplete(metadata) => Some(metadata),
+                LocalTimelineInitStatus::NeedsSync => None,
+            }),
+        )
     } else {
         info!("No remote storage configured, skipping storage sync, considering all local timelines with correct metadata files enabled");
-        (RemoteIndex::default(), local_tenant_files)
+        (
+            RemoteIndex::default(),
+            local_tenant_files.filter_map(|(metadata, _)| Some(metadata)),
+        )
     };
+
     attach_local_tenants(conf, &remote_index, tenants_to_attach);
 
     Ok(remote_index)
@@ -137,12 +117,18 @@ pub fn init_tenant_mgr(
 pub fn attach_local_tenants(
     conf: &'static PageServerConf,
     remote_index: &RemoteIndex,
-    tenants_to_attach: HashMap<TenantId, TenantAttachData>,
+    tenants_to_attach: TenantTimelineValues<TimelineMetadata>,
 ) {
     let _entered = info_span!("attach_local_tenants").entered();
-    let number_of_tenants = tenants_to_attach.len();
+    let number_of_tenants = tenants_to_attach.0.len();
+
+    for (tenant_id, local_timelines) in tenants_to_attach.0 {
+        info!(
+            "Attaching {} timelines for {tenant_id}",
+            local_timelines.len()
+        );
+        debug!("Timelines to attach: {local_timelines:?}");
 
-    for (tenant_id, local_timelines) in tenants_to_attach {
         let mut tenants_accessor = tenants_state::write_tenants();
         let tenant = match tenants_accessor.entry(tenant_id) {
             hash_map::Entry::Occupied(o) => {
@@ -151,55 +137,25 @@ pub fn attach_local_tenants(
             }
             hash_map::Entry::Vacant(v) => {
                 info!("Tenant {tenant_id} was not found in pageserver's memory, loading it");
-                let tenant = Arc::new(Tenant::new(
-                    conf,
-                    TenantConfOpt::default(),
-                    Arc::new(PostgresRedoManager::new(conf, tenant_id)),
-                    tenant_id,
-                    remote_index.clone(),
-                    conf.remote_storage_config.is_some(),
-                ));
-                match local_timelines {
-                    TenantAttachData::Broken(_) => {
-                        tenant.set_state(TenantState::Broken);
-                    }
-                    TenantAttachData::Ready(_) => {
-                        match Tenant::load_tenant_config(conf, tenant_id) {
-                            Ok(tenant_conf) => {
-                                tenant.update_tenant_config(tenant_conf);
-                                tenant.activate(false);
-                            }
-                            Err(e) => {
-                                error!("Failed to read config for tenant {tenant_id}, disabling tenant: {e:?}");
-                                tenant.set_state(TenantState::Broken);
-                            }
-                        };
-                    }
-                }
+                let tenant = load_local_tenant(conf, tenant_id, remote_index);
                 v.insert(Arc::clone(&tenant));
                 tenant
             }
         };
         drop(tenants_accessor);
-        match local_timelines {
-            TenantAttachData::Broken(e) => warn!("{}", e),
-            TenantAttachData::Ready(ref timelines) => {
-                info!("Attaching {} timelines for {tenant_id}", timelines.len());
-                debug!("Timelines to attach: {local_timelines:?}");
-                let has_timelines = !timelines.is_empty();
-                let timelines_to_attach = timelines
-                    .iter()
-                    .map(|(&k, v)| (k, v.metadata().to_owned()))
-                    .collect();
-                match tenant.init_attach_timelines(timelines_to_attach) {
-                    Ok(()) => {
-                        info!("successfully loaded local timelines for tenant {tenant_id}");
-                        tenant.activate(has_timelines);
-                    }
-                    Err(e) => {
-                        error!("Failed to attach tenant timelines: {e:?}");
-                        tenant.set_state(TenantState::Broken);
-                    }
+
+        if tenant.current_state() == TenantState::Broken {
+            warn!("Skipping timeline load for broken tenant {tenant_id}")
+        } else {
+            let has_timelines = !local_timelines.is_empty();
+            match tenant.init_attach_timelines(local_timelines) {
+                Ok(()) => {
+                    info!("successfully loaded local timelines for tenant {tenant_id}");
+                    tenant.activate(has_timelines);
+                }
+                Err(e) => {
+                    error!("Failed to attach tenant timelines: {e:?}");
+                    tenant.set_state(TenantState::Broken);
                 }
             }
         }
@@ -208,6 +164,44 @@ pub fn attach_local_tenants(
     info!("Processed {number_of_tenants} local tenants during attach")
 }
 
+fn load_local_tenant(
+    conf: &'static PageServerConf,
+    tenant_id: TenantId,
+    remote_index: &RemoteIndex,
+) -> Arc<Tenant> {
+    let tenant = Arc::new(Tenant::new(
+        conf,
+        TenantConfOpt::default(),
+        Arc::new(PostgresRedoManager::new(conf, tenant_id)),
+        tenant_id,
+        remote_index.clone(),
+        conf.remote_storage_config.is_some(),
+    ));
+
+    let tenant_timelines_dir = conf.timelines_path(&tenant_id);
+    if !tenant_timelines_dir.is_dir() {
+        error!(
+            "Tenant {} has no timelines directory at {}",
+            tenant_id,
+            tenant_timelines_dir.display()
+        );
+        tenant.set_state(TenantState::Broken);
+    } else {
+        match Tenant::load_tenant_config(conf, tenant_id) {
+            Ok(tenant_conf) => {
+                tenant.update_tenant_config(tenant_conf);
+                tenant.activate(false);
+            }
+            Err(e) => {
+                error!("Failed to read config for tenant {tenant_id}, disabling tenant: {e:?}");
+                tenant.set_state(TenantState::Broken);
+            }
+        }
+    }
+
+    tenant
+}
+
 ///
 /// Shut down all tenants. This runs as part of pageserver shutdown.
 ///
@@ -265,98 +259,58 @@ fn create_tenant_files(
         temporary_tenant_dir.display()
     );
 
+    let temporary_tenant_timelines_dir = rebase_directory(
+        &conf.timelines_path(&tenant_id),
+        &target_tenant_directory,
+        &temporary_tenant_dir,
+    )?;
+    let temporary_tenant_config_path = rebase_directory(
+        &conf.tenant_config_path(tenant_id),
+        &target_tenant_directory,
+        &temporary_tenant_dir,
+    )?;
+
     // top-level dir may exist if we are creating it through CLI
-    crashsafe::create_dir_all(&temporary_tenant_dir).with_context(|| {
+    crashsafe_dir::create_dir_all(&temporary_tenant_dir).with_context(|| {
         format!(
             "could not create temporary tenant directory {}",
             temporary_tenant_dir.display()
         )
     })?;
-
-    let creation_result = try_create_target_tenant_dir(
-        conf,
-        tenant_conf,
-        tenant_id,
-        &temporary_tenant_dir,
-        &target_tenant_directory,
-    );
-
-    if creation_result.is_err() {
-        error!("Failed to create directory structure for tenant {tenant_id}, cleaning tmp data");
-        if let Err(e) = fs::remove_dir_all(&temporary_tenant_dir) {
-            error!("Failed to remove temporary tenant directory {temporary_tenant_dir:?}: {e}")
-        } else if let Err(e) = crashsafe::fsync(&temporary_tenant_dir) {
-            error!(
-                "Failed to fsync removed temporary tenant directory {temporary_tenant_dir:?}: {e}"
-            )
-        }
-    }
-
-    creation_result
-}
-
-fn try_create_target_tenant_dir(
-    conf: &'static PageServerConf,
-    tenant_conf: TenantConfOpt,
-    tenant_id: TenantId,
-    temporary_tenant_dir: &Path,
-    target_tenant_directory: &Path,
-) -> Result<(), anyhow::Error> {
-    let temporary_tenant_timelines_dir = rebase_directory(
-        &conf.timelines_path(&tenant_id),
-        target_tenant_directory,
-        temporary_tenant_dir,
-    )
-    .with_context(|| format!("Failed to resolve tenant {tenant_id} temporary timelines dir"))?;
-    let temporary_tenant_config_path = rebase_directory(
-        &conf.tenant_config_path(tenant_id),
-        target_tenant_directory,
-        temporary_tenant_dir,
-    )
-    .with_context(|| format!("Failed to resolve tenant {tenant_id} temporary config path"))?;
-
-    Tenant::persist_tenant_config(&temporary_tenant_config_path, tenant_conf, true).with_context(
-        || {
-            format!(
-                "Failed to write tenant {} config to {}",
-                tenant_id,
-                temporary_tenant_config_path.display()
-            )
-        },
-    )?;
-    crashsafe::create_dir(&temporary_tenant_timelines_dir).with_context(|| {
+    // first, create a config in the top-level temp directory, fsync the file
+    Tenant::persist_tenant_config(&temporary_tenant_config_path, tenant_conf, true)?;
+    // then, create a subdirectory in the top-level temp directory, fsynced
+    crashsafe_dir::create_dir(&temporary_tenant_timelines_dir).with_context(|| {
         format!(
-            "could not create tenant {} temporary timelines directory {}",
-            tenant_id,
+            "could not create temporary tenant timelines directory {}",
             temporary_tenant_timelines_dir.display()
         )
     })?;
+
     fail::fail_point!("tenant-creation-before-tmp-rename", |_| {
         anyhow::bail!("failpoint tenant-creation-before-tmp-rename");
     });
 
-    fs::rename(&temporary_tenant_dir, target_tenant_directory).with_context(|| {
+    // move-rename tmp directory with all files synced into a permanent directory, fsync its parent
+    fs::rename(&temporary_tenant_dir, &target_tenant_directory).with_context(|| {
         format!(
-            "failed to move tenant {} temporary directory {} into the permanent one {}",
-            tenant_id,
+            "failed to move temporary tenant directory {} into the permanent one {}",
             temporary_tenant_dir.display(),
             target_tenant_directory.display()
         )
     })?;
     let target_dir_parent = target_tenant_directory.parent().with_context(|| {
         format!(
-            "Failed to get tenant {} dir parent for {}",
-            tenant_id,
+            "Failed to get tenant dir parent for {}",
             target_tenant_directory.display()
         )
     })?;
-    crashsafe::fsync(target_dir_parent).with_context(|| {
-        format!(
-            "Failed to fsync renamed directory's parent {} for tenant {}",
-            target_dir_parent.display(),
-            tenant_id,
-        )
-    })?;
+    fs::File::open(target_dir_parent)?.sync_all()?;
+
+    info!(
+        "created tenant directory structure in {}",
+        target_tenant_directory.display()
+    );
 
     Ok(())
 }
@@ -521,21 +475,16 @@ pub fn list_tenant_info(remote_index: &RemoteTimelineIndex) -> Vec<TenantInfo> {
         .collect()
 }
 
-#[derive(Debug)]
-pub enum TenantAttachData {
-    Ready(HashMap<TimelineId, TimelineLocalFiles>),
-    Broken(anyhow::Error),
-}
 /// Attempts to collect information about all tenant and timelines, existing on the local FS.
 /// If finds any, deletes all temporary files and directories, created before. Also removes empty directories,
 /// that may appear due to such removals.
 /// Does not fail on particular timeline or tenant collection errors, rather logging them and ignoring the entities.
 fn local_tenant_timeline_files(
     config: &'static PageServerConf,
-) -> anyhow::Result<HashMap<TenantId, TenantAttachData>> {
+) -> anyhow::Result<TenantTimelineValues<(TimelineMetadata, HashSet<PathBuf>)>> {
     let _entered = info_span!("local_tenant_timeline_files").entered();
 
-    let mut local_tenant_timeline_files = HashMap::new();
+    let mut local_tenant_timeline_files = TenantTimelineValues::new();
     let tenants_dir = config.tenants_path();
     for tenants_dir_entry in fs::read_dir(&tenants_dir)
         .with_context(|| format!("Failed to list tenants dir {}", tenants_dir.display()))?
@@ -557,31 +506,19 @@ fn local_tenant_timeline_files(
                     }
                 } else {
                     match collect_timelines_for_tenant(config, &tenant_dir_path) {
-                        Ok((tenant_id, TenantAttachData::Broken(e))) => {
-                            local_tenant_timeline_files.entry(tenant_id).or_insert(TenantAttachData::Broken(e));
-                        },
-                        Ok((tenant_id, TenantAttachData::Ready(collected_files))) => {
+                        Ok((tenant_id, collected_files)) => {
                             if collected_files.is_empty() {
                                 match remove_if_empty(&tenant_dir_path) {
                                     Ok(true) => info!("Removed empty tenant directory {}", tenant_dir_path.display()),
                                     Ok(false) => {
                                         // insert empty timeline entry: it has some non-temporary files inside that we cannot remove
                                         // so make obvious for HTTP API callers, that something exists there and try to load the tenant
-                                        let _ = local_tenant_timeline_files.entry(tenant_id).or_insert_with(|| TenantAttachData::Ready(HashMap::new()));
+                                        let _ = local_tenant_timeline_files.0.entry(tenant_id).or_default();
                                     },
                                     Err(e) => error!("Failed to remove empty tenant directory: {e:?}"),
                                 }
                             } else {
-                                match local_tenant_timeline_files.entry(tenant_id) {
-                                    hash_map::Entry::Vacant(entry) => {
-                                        entry.insert(TenantAttachData::Ready(collected_files));
-                                    }
-                                    hash_map::Entry::Occupied(entry) =>{
-                                        if let TenantAttachData::Ready(old_timelines) = entry.into_mut() {
-                                            old_timelines.extend(collected_files);
-                                        }
-                                    },
-                                }
+                                local_tenant_timeline_files.0.entry(tenant_id).or_default().extend(collected_files.into_iter())
                             }
                         },
                         Err(e) => error!(
@@ -604,7 +541,7 @@ fn local_tenant_timeline_files(
 
     info!(
         "Collected files for {} tenants",
-        local_tenant_timeline_files.len(),
+        local_tenant_timeline_files.0.len()
     );
     Ok(local_tenant_timeline_files)
 }
@@ -642,19 +579,14 @@ fn is_temporary(path: &Path) -> bool {
     }
 }
 
-fn is_uninit_mark(path: &Path) -> bool {
-    match path.file_name() {
-        Some(name) => name
-            .to_string_lossy()
-            .ends_with(TIMELINE_UNINIT_MARK_SUFFIX),
-        None => false,
-    }
-}
-
+#[allow(clippy::type_complexity)]
 fn collect_timelines_for_tenant(
     config: &'static PageServerConf,
     tenant_path: &Path,
-) -> anyhow::Result<(TenantId, TenantAttachData)> {
+) -> anyhow::Result<(
+    TenantId,
+    HashMap<TimelineId, (TimelineMetadata, HashSet<PathBuf>)>,
+)> {
     let tenant_id = tenant_path
         .file_name()
         .and_then(OsStr::to_str)
@@ -663,17 +595,6 @@ fn collect_timelines_for_tenant(
         .context("Could not parse tenant id out of the tenant dir name")?;
     let timelines_dir = config.timelines_path(&tenant_id);
 
-    if !timelines_dir.as_path().is_dir() {
-        return Ok((
-            tenant_id,
-            TenantAttachData::Broken(anyhow::anyhow!(
-                "Tenant {} has no timelines directory at {}",
-                tenant_id,
-                timelines_dir.display()
-            )),
-        ));
-    }
-
     let mut tenant_timelines = HashMap::new();
     for timelines_dir_entry in fs::read_dir(&timelines_dir)
         .with_context(|| format!("Failed to list timelines dir entry for tenant {tenant_id}"))?
@@ -693,74 +614,25 @@ fn collect_timelines_for_tenant(
                             e
                         );
                     }
-                } else if is_uninit_mark(&timeline_dir) {
-                    let timeline_uninit_mark_file = &timeline_dir;
-                    info!(
-                        "Found an uninit mark file {}, removing the timeline and its uninit mark",
-                        timeline_uninit_mark_file.display()
-                    );
-                    let timeline_id = timeline_uninit_mark_file
-                        .file_stem()
-                        .and_then(OsStr::to_str)
-                        .unwrap_or_default()
-                        .parse::<TimelineId>()
-                        .with_context(|| {
-                            format!(
-                                "Could not parse timeline id out of the timeline uninit mark name {}",
-                                timeline_uninit_mark_file.display()
-                            )
-                        })?;
-                    let timeline_dir = config.timeline_path(&timeline_id, &tenant_id);
-                    if let Err(e) =
-                        remove_timeline_and_uninit_mark(&timeline_dir, timeline_uninit_mark_file)
-                    {
-                        error!("Failed to clean up uninit marked timeline: {e:?}");
-                    }
                 } else {
-                    let timeline_id = timeline_dir
-                        .file_name()
-                        .and_then(OsStr::to_str)
-                        .unwrap_or_default()
-                        .parse::<TimelineId>()
-                        .with_context(|| {
-                            format!(
-                                "Could not parse timeline id out of the timeline dir name {}",
-                                timeline_dir.display()
-                            )
-                        })?;
-                    let timeline_uninit_mark_file =
-                        config.timeline_uninit_mark_file_path(tenant_id, timeline_id);
-                    if timeline_uninit_mark_file.exists() {
-                        info!("Found an uninit mark file for timeline {tenant_id}/{timeline_id}, removing the timeline and its uninit mark");
-                        if let Err(e) = remove_timeline_and_uninit_mark(
-                            &timeline_dir,
-                            &timeline_uninit_mark_file,
-                        ) {
-                            error!("Failed to clean up uninit marked timeline: {e:?}");
+                    match collect_timeline_files(&timeline_dir) {
+                        Ok((timeline_id, metadata, timeline_files)) => {
+                            tenant_timelines.insert(timeline_id, (metadata, timeline_files));
                         }
-                    } else {
-                        match collect_timeline_files(&timeline_dir) {
-                            Ok((metadata, timeline_files)) => {
-                                tenant_timelines.insert(
-                                    timeline_id,
-                                    TimelineLocalFiles::collected(metadata, timeline_files),
-                                );
-                            }
-                            Err(e) => {
-                                error!(
-                                    "Failed to process timeline dir contents at '{}', reason: {:?}",
-                                    timeline_dir.display(),
-                                    e
-                                );
-                                match remove_if_empty(&timeline_dir) {
-                                    Ok(true) => info!(
-                                        "Removed empty timeline directory {}",
-                                        timeline_dir.display()
-                                    ),
-                                    Ok(false) => (),
-                                    Err(e) => {
-                                        error!("Failed to remove empty timeline directory: {e:?}")
-                                    }
+                        Err(e) => {
+                            error!(
+                                "Failed to process timeline dir contents at '{}', reason: {:?}",
+                                timeline_dir.display(),
+                                e
+                            );
+                            match remove_if_empty(&timeline_dir) {
+                                Ok(true) => info!(
+                                    "Removed empty timeline directory {}",
+                                    timeline_dir.display()
+                                ),
+                                Ok(false) => (),
+                                Err(e) => {
+                                    error!("Failed to remove empty timeline directory: {e:?}")
                                 }
                             }
                         }
@@ -780,51 +652,28 @@ fn collect_timelines_for_tenant(
         debug!("Tenant {tenant_id} has no timelines loaded");
     }
 
-    Ok((tenant_id, TenantAttachData::Ready(tenant_timelines)))
-}
-
-fn remove_timeline_and_uninit_mark(timeline_dir: &Path, uninit_mark: &Path) -> anyhow::Result<()> {
-    fs::remove_dir_all(&timeline_dir)
-        .or_else(|e| {
-            if e.kind() == std::io::ErrorKind::NotFound {
-                // we can leave the uninit mark without a timeline dir,
-                // just remove the mark then
-                Ok(())
-            } else {
-                Err(e)
-            }
-        })
-        .with_context(|| {
-            format!(
-                "Failed to remove unit marked timeline directory {}",
-                timeline_dir.display()
-            )
-        })?;
-    fs::remove_file(&uninit_mark).with_context(|| {
-        format!(
-            "Failed to remove timeline uninit mark file {}",
-            uninit_mark.display()
-        )
-    })?;
-
-    Ok(())
+    Ok((tenant_id, tenant_timelines))
 }
 
 // discover timeline files and extract timeline metadata
 //  NOTE: ephemeral files are excluded from the list
 fn collect_timeline_files(
     timeline_dir: &Path,
-) -> anyhow::Result<(TimelineMetadata, HashMap<PathBuf, LayerFileMetadata>)> {
-    let mut timeline_files = HashMap::new();
+) -> anyhow::Result<(TimelineId, TimelineMetadata, HashSet<PathBuf>)> {
+    let mut timeline_files = HashSet::new();
     let mut timeline_metadata_path = None;
 
+    let timeline_id = timeline_dir
+        .file_name()
+        .and_then(OsStr::to_str)
+        .unwrap_or_default()
+        .parse::<TimelineId>()
+        .context("Could not parse timeline id out of the timeline dir name")?;
     let timeline_dir_entries =
         fs::read_dir(&timeline_dir).context("Failed to list timeline dir contents")?;
     for entry in timeline_dir_entries {
         let entry_path = entry.context("Failed to list timeline dir entry")?.path();
-        let metadata = entry_path.metadata()?;
-
-        if metadata.is_file() {
+        if entry_path.is_file() {
             if entry_path.file_name().and_then(OsStr::to_str) == Some(METADATA_FILE_NAME) {
                 timeline_metadata_path = Some(entry_path);
             } else if is_ephemeral_file(&entry_path.file_name().unwrap().to_string_lossy()) {
@@ -839,8 +688,7 @@ fn collect_timeline_files(
                     )
                 })?;
             } else {
-                let layer_metadata = LayerFileMetadata::new(metadata.len());
-                timeline_files.insert(entry_path, layer_metadata);
+                timeline_files.insert(entry_path);
             }
         }
     }
@@ -866,5 +714,5 @@ fn collect_timeline_files(
         "Timeline has no ancestor and no layer files"
     );
 
-    Ok((metadata, timeline_files))
+    Ok((timeline_id, metadata, timeline_files))
 }

pageserver/src/tenant_tasks.rs

@@ -70,10 +70,8 @@ async fn compaction_loop(tenant_id: TenantId) {
             // Run compaction
             let mut sleep_duration = tenant.get_compaction_period();
             if let Err(e) = tenant.compaction_iteration() {
+                error!("Compaction failed, retrying: {e:#}");
                 sleep_duration = wait_duration;
-                error!("Compaction failed, retrying in {:?}: {e:#}", sleep_duration);
-                #[cfg(feature = "testing")]
-                std::process::abort();
             }
 
             // Sleep
@@ -121,10 +119,8 @@ async fn gc_loop(tenant_id: TenantId) {
             if gc_horizon > 0 {
                 if let Err(e) = tenant.gc_iteration(None, gc_horizon, tenant.get_pitr_interval(), false)
                 {
+                    error!("Gc failed, retrying: {e:#}");
                     sleep_duration = wait_duration;
-                    error!("Gc failed, retrying in {:?}: {e:#}", sleep_duration);
-                    #[cfg(feature = "testing")]
-                    std::process::abort();
                 }
             }
 
@@ -175,7 +171,7 @@ async fn wait_for_active_tenant(
                         }
                         state => {
                             debug!("Not running the task loop, tenant is not active with background jobs enabled: {state:?}");
-                            continue;
+                            tokio::time::sleep(wait).await;
                         }
                     }
                 }

pageserver/src/walingest.rs

@@ -31,10 +31,9 @@ use bytes::{Buf, Bytes, BytesMut};
 use tracing::*;
 
 use crate::pgdatadir_mapping::*;
+use crate::reltag::{RelTag, SlruKind};
 use crate::tenant::Timeline;
 use crate::walrecord::*;
-use crate::ZERO_PAGE;
-use pageserver_api::reltag::{RelTag, SlruKind};
 use postgres_ffi::pg_constants;
 use postgres_ffi::relfile_utils::{FSM_FORKNUM, MAIN_FORKNUM, VISIBILITYMAP_FORKNUM};
 use postgres_ffi::v14::nonrelfile_utils::mx_offset_to_member_segment;
@@ -44,6 +43,8 @@ use postgres_ffi::TransactionId;
 use postgres_ffi::BLCKSZ;
 use utils::lsn::Lsn;
 
+static ZERO_PAGE: Bytes = Bytes::from_static(&[0u8; 8192]);
+
 pub struct WalIngest<'a> {
     timeline: &'a Timeline,