Add tokio-console tracing as optional feature.

Which ought to replace neon_broker. This patch only adds the binary and adjusts
Dockerfile to include it; subsequent ones will add deploy of helm chart and the
actual replacement.

It is a simple and fast pub-sub message bus. In this patch only safekeeper
message is supported, but others can be easily added.

Compilation now requires protoc to be installed. Installing protobuf-compiler
package is fine for Debian/Ubuntu.

ref
https://github.com/neondatabase/neon/pull/2733
https://github.com/neondatabase/neon/issues/2394

.dockerignore

@@ -5,6 +5,7 @@
 !Cargo.lock
 !Makefile
 
+!broker/
 !.cargo/
 !.config/
 !control_plane/

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

@@ -55,6 +55,22 @@ runs:
         name: neon-${{ runner.os }}-${{ inputs.build_type }}-artifact
         path: /tmp/neon
 
+    - name: Download Neon binaries for the previous release
+      if: inputs.build_type != 'remote'
+      uses: ./.github/actions/download
+      with:
+        name: neon-${{ runner.os }}-${{ inputs.build_type }}-artifact
+        path: /tmp/neon-previous
+        prefix: latest
+
+    - name: Download compatibility snapshot for Postgres 14
+      if: inputs.build_type != 'remote'
+      uses: ./.github/actions/download
+      with:
+        name: compatibility-snapshot-${{ inputs.build_type }}-pg14
+        path: /tmp/compatibility_snapshot_pg14
+        prefix: latest
+
     - name: Checkout
       if: inputs.needs_postgres_source == 'true'
       uses: actions/checkout@v3
@@ -76,10 +92,15 @@ runs:
     - name: Run pytest
       env:
         NEON_BIN: /tmp/neon/bin
+        COMPATIBILITY_NEON_BIN: /tmp/neon-previous/bin
+        COMPATIBILITY_POSTGRES_DISTRIB_DIR: /tmp/neon-previous/pg_install
         TEST_OUTPUT: /tmp/test_output
         BUILD_TYPE: ${{ inputs.build_type }}
         AWS_ACCESS_KEY_ID: ${{ inputs.real_s3_access_key_id }}
         AWS_SECRET_ACCESS_KEY: ${{ inputs.real_s3_secret_access_key }}
+        COMPATIBILITY_SNAPSHOT_DIR: /tmp/compatibility_snapshot_pg14
+        ALLOW_BACKWARD_COMPATIBILITY_BREAKAGE: contains(github.event.pull_request.labels.*.name, 'backward compatibility breakage')
+        ALLOW_FORWARD_COMPATIBILITY_BREAKAGE: contains(github.event.pull_request.labels.*.name, 'forward compatibility breakage')
       shell: bash -euxo pipefail {0}
       run: |
         # PLATFORM will be embedded in the perf test report
@@ -102,7 +123,12 @@ runs:
           exit 1
         fi
         if [[ "${{ inputs.run_in_parallel }}" == "true" ]]; then
+          # -n4 uses four processes to run tests via pytest-xdist
           EXTRA_PARAMS="-n4 $EXTRA_PARAMS"
+
+          # --dist=loadgroup points tests marked with @pytest.mark.xdist_group
+          # to the same worker to make @pytest.mark.order work with xdist
+          EXTRA_PARAMS="--dist=loadgroup $EXTRA_PARAMS"
         fi
 
         if [[ "${{ inputs.run_with_real_s3 }}" == "true" ]]; then
@@ -137,9 +163,9 @@ runs:
         # --verbose prints name of each test (helpful when there are
         # multiple tests in one file)
         # -rA prints summary in the end
-        # -n4 uses four processes to run tests via pytest-xdist
         # -s is not used to prevent pytest from capturing output, because tests are running
         # in parallel and logs are mixed between different tests
+        #
         mkdir -p $TEST_OUTPUT/allure/results
         "${cov_prefix[@]}" ./scripts/pytest \
           --junitxml=$TEST_OUTPUT/junit.xml \
@@ -154,8 +180,17 @@ 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_create_snapshot) and directory that the test creates (compatibility_snapshot_pg14), keep the path in sync with the test
+        path: /tmp/test_output/test_create_snapshot/compatibility_snapshot_pg14/
+        prefix: latest
+
     - name: Create Allure report
-      if: always()
+      if: success() || failure()
       uses: ./.github/actions/allure-report
       with:
         action: store

.github/ansible/deploy.yaml

@@ -1,7 +1,7 @@
 - name: Upload Neon binaries
   hosts: storage
   gather_facts: False
-  remote_user: admin
+  remote_user: "{{ remote_user }}"
 
   tasks:
 
@@ -36,7 +36,7 @@
 - name: Deploy pageserver
   hosts: pageservers
   gather_facts: False
-  remote_user: admin
+  remote_user: "{{ remote_user }}"
 
   tasks:
 
@@ -124,7 +124,7 @@
 - name: Deploy safekeeper
   hosts: safekeepers
   gather_facts: False
-  remote_user: admin
+  remote_user: "{{ remote_user }}"
 
   tasks:
 

.github/ansible/get_binaries.sh

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

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

@@ -3,7 +3,6 @@ storage:
     bucket_name: neon-storage-ireland
     bucket_region: eu-west-1
     console_mgmt_base_url: http://neon-stress-console.local
-    env_name: neon-stress
     etcd_endpoints: neon-stress-etcd.local:2379
     safekeeper_enable_s3_offload: 'false'
     pageserver_config_stub:
@@ -12,19 +11,21 @@ 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: 1
+          console_region_id: aws-eu-west-1
         neon-stress-ps-2:
-          console_region_id: 1
+          console_region_id: aws-eu-west-1
     safekeepers:
       hosts:
         neon-stress-sk-1:
-          console_region_id: 1
+          console_region_id: aws-eu-west-1
         neon-stress-sk-2:
-          console_region_id: 1
+          console_region_id: aws-eu-west-1
         neon-stress-sk-3:
-          console_region_id: 1
+          console_region_id: aws-eu-west-1

.github/ansible/prod.ap-southeast-1.hosts.yaml

@@ -0,0 +1,35 @@
+storage:
+  vars:
+    bucket_name: neon-prod-storage-ap-southeast-1
+    bucket_region: ap-southeast-1
+    console_mgmt_base_url: http://console-release.local
+    etcd_endpoints: etcd-0.ap-southeast-1.aws.neon.tech: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: ap-southeast-1
+    ansible_aws_ssm_bucket_name: neon-prod-storage-ap-southeast-1
+    console_region_id: aws-ap-southeast-1
+
+  children:
+    pageservers:
+      hosts:
+        pageserver-0.ap-southeast-1.aws.neon.tech:
+          ansible_host:  i-064de8ea28bdb495b
+        pageserver-1.ap-southeast-1.aws.neon.tech:
+          ansible_host:  i-0b180defcaeeb6b93
+
+    safekeepers:
+      hosts:
+        safekeeper-0.ap-southeast-1.aws.neon.tech:
+          ansible_host:  i-0d6f1dc5161eef894
+        safekeeper-1.ap-southeast-1.aws.neon.tech:
+          ansible_host:  i-0e338adda8eb2d19f
+        safekeeper-2.ap-southeast-1.aws.neon.tech:
+          ansible_host:  i-04fb63634e4679eb9

.github/ansible/prod.eu-central-1.hosts.yaml

@@ -0,0 +1,35 @@
+storage:
+  vars:
+    bucket_name: neon-prod-storage-eu-central-1
+    bucket_region: eu-central-1
+    console_mgmt_base_url: http://console-release.local
+    etcd_endpoints: etcd-0.eu-central-1.aws.neon.tech:2379
+    pageserver_config_stub:
+      pg_distrib_dir: /usr/local
+      remote_storage:
+        bucket_name: "{{ bucket_name }}"
+        bucket_region: "{{ bucket_region }}"
+        prefix_in_bucket: "pageserver/v1"
+    safekeeper_s3_prefix: safekeeper/v1/wal
+    hostname_suffix: ""
+    remote_user: ssm-user
+    ansible_aws_ssm_region: eu-central-1
+    ansible_aws_ssm_bucket_name: neon-prod-storage-eu-central-1
+    console_region_id: aws-eu-central-1
+
+  children:
+    pageservers:
+      hosts:
+        pageserver-0.eu-central-1.aws.neon.tech:
+          ansible_host:  i-0cd8d316ecbb715be
+        pageserver-1.eu-central-1.aws.neon.tech:
+          ansible_host:  i-090044ed3d383fef0
+
+    safekeepers:
+      hosts:
+        safekeeper-0.eu-central-1.aws.neon.tech:
+          ansible_host:  i-0b238612d2318a050
+        safekeeper-1.eu-central-1.aws.neon.tech:
+          ansible_host:  i-07b9c45e5c2637cd4
+        safekeeper-2.eu-central-1.aws.neon.tech:
+          ansible_host:  i-020257302c3c93d88

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

@@ -0,0 +1,36 @@
+storage:
+  vars:
+    bucket_name: neon-prod-storage-us-east-2
+    bucket_region: us-east-2
+    console_mgmt_base_url: http://console-release.local
+    etcd_endpoints: etcd-0.us-east-2.aws.neon.tech: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
+    ansible_aws_ssm_bucket_name: neon-prod-storage-us-east-2
+    console_region_id: aws-us-east-2
+
+  children:
+    pageservers:
+      hosts:
+        pageserver-0.us-east-2.aws.neon.tech:
+          ansible_host:  i-062227ba7f119eb8c
+        pageserver-1.us-east-2.aws.neon.tech:
+          ansible_host:  i-0b3ec0afab5968938
+
+    safekeepers:
+      hosts:
+        safekeeper-0.us-east-2.aws.neon.tech:
+          ansible_host:  i-0e94224750c57d346
+        safekeeper-1.us-east-2.aws.neon.tech:
+          ansible_host:  i-06d113fb73bfddeb0
+        safekeeper-2.us-east-2.aws.neon.tech:
+          ansible_host:  i-09f66c8e04afff2e8
+          

.github/ansible/production.hosts.yaml

@@ -1,7 +1,6 @@
 ---
 storage:
   vars:
-    env_name: prod-1
     console_mgmt_base_url: http://console-release.local
     bucket_name: zenith-storage-oregon
     bucket_region: us-west-2
@@ -12,20 +11,23 @@ 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: 1
+          console_region_id: aws-us-west-2
         zenith-1-ps-3:
-          console_region_id: 1
+          console_region_id: aws-us-west-2
 
     safekeepers:
       hosts:
         zenith-1-sk-1:
-          console_region_id: 1
+          console_region_id: aws-us-west-2
         zenith-1-sk-2:
-          console_region_id: 1
+          console_region_id: aws-us-west-2
         zenith-1-sk-3:
-          console_region_id: 1
+          console_region_id: aws-us-west-2

.github/ansible/scripts/init_pageserver.sh

@@ -12,18 +12,19 @@ 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
+  "http_port": 9898,
+  "active": false
 }
 EOF
 
 # check if pageserver already registered or not
-if ! curl -sf -X PATCH -d '{}' {{ console_mgmt_base_url }}/api/v1/pageservers/${INSTANCE_ID} -o /dev/null; then
+if ! curl -sf -H "Authorization: Bearer {{ CONSOLE_API_TOKEN }}" {{ console_mgmt_base_url }}/management/api/v2/pageservers/${INSTANCE_ID} -o /dev/null; then
 
     # not registered, so register it now
-    ID=$(curl -sf -X POST {{ console_mgmt_base_url }}/api/v1/pageservers -d@/tmp/payload | jq -r '.ID')
+    ID=$(curl -sf -X POST -H "Authorization: Bearer {{ CONSOLE_API_TOKEN }}" {{ console_mgmt_base_url }}/management/api/v2/pageservers -d@/tmp/payload | jq -r '.id')
 
     # 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": "${AZ_ID}"
+  "availability_zone_id": "${AZ_ID}",
+  "active": false
 }
 EOF
 
 # check if safekeeper already registered or not
-if ! curl -sf -X PATCH -d '{}' {{ console_mgmt_base_url }}/api/v1/safekeepers/${INSTANCE_ID} -o /dev/null; then
+if ! curl -sf -H "Authorization: Bearer {{ CONSOLE_API_TOKEN }}" {{ console_mgmt_base_url }}/management/api/v2/safekeepers/${INSTANCE_ID} -o /dev/null; then
 
     # not registered, so register it now
-    ID=$(curl -sf -X POST {{ console_mgmt_base_url }}/api/v1/safekeepers -d@/tmp/payload | jq -r '.ID')
-
+    ID=$(curl -sf -X POST -H "Authorization: Bearer {{ CONSOLE_API_TOKEN }}" {{ console_mgmt_base_url }}/management/api/v2/safekeepers -d@/tmp/payload | jq -r '.id')
     # init safekeeper
     sudo -u safekeeper /usr/local/bin/safekeeper --id ${ID} --init -D /storage/safekeeper/data
 fi

.github/ansible/ssm_config

@@ -0,0 +1,2 @@
+ansible_connection: aws_ssm
+ansible_python_interpreter: /usr/bin/python3

.github/ansible/staging.hosts.yaml

@@ -3,7 +3,6 @@ storage:
     bucket_name: zenith-staging-storage-us-east-1
     bucket_region: us-east-1
     console_mgmt_base_url: http://console-staging.local
-    env_name: us-stage
     etcd_endpoints: zenith-us-stage-etcd.local:2379
     pageserver_config_stub:
       pg_distrib_dir: /usr/local
@@ -11,30 +10,25 @@ 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: 27
+          console_region_id: aws-us-east-1
         zenith-us-stage-ps-3:
-          console_region_id: 27
+          console_region_id: aws-us-east-1
         zenith-us-stage-ps-4:
-          console_region_id: 27
-        zenith-us-stage-test-ps-1:
-          console_region_id: 28
+          console_region_id: aws-us-east-1
 
     safekeepers:
       hosts:
         zenith-us-stage-sk-4:
-          console_region_id: 27
+          console_region_id: aws-us-east-1
         zenith-us-stage-sk-5:
-          console_region_id: 27
+          console_region_id: aws-us-east-1
         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
+          console_region_id: aws-us-east-1

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

@@ -0,0 +1,33 @@
+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
+    ansible_aws_ssm_bucket_name: neon-staging-storage-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=Zenith pageserver
+Description=Neon pageserver
 After=network.target auditd.service
 
 [Service]

.github/ansible/systemd/safekeeper.service

@@ -1,12 +1,12 @@
 [Unit]
-Description=Zenith safekeeper
+Description=Neon 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 }}.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"}'
+ExecStart=/usr/local/bin/safekeeper -l {{ inventory_hostname }}{{ hostname_suffix }}:6500 --listen-http {{ inventory_hostname }}{{ hostname_suffix }}:7676 -D /storage/safekeeper/data --broker-endpoints={{ etcd_endpoints }} --remote-storage='{bucket_name="{{bucket_name}}", bucket_region="{{bucket_region}}", prefix_in_bucket="{{ safekeeper_s3_prefix }}"}'
 ExecReload=/bin/kill -HUP $MAINPID
 KillMode=mixed
 KillSignal=SIGINT

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

@@ -0,0 +1,31 @@
+# Helm chart values for neon-proxy-scram.
+# This is a YAML-formatted file.
+
+image:
+  repository: neondatabase/neon
+
+settings:
+  authBackend: "console"
+  authEndpoint: "http://console-staging.local/management/api/v2"
+  domain: "*.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/helm-values/prod-ap-southeast-1-epsilon.neon-proxy-scram.yaml

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

.github/helm-values/prod-eu-central-1-gamma.neon-proxy-scram.yaml

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

.github/helm-values/prod-us-east-2-delta.neon-proxy-scram.yaml

@@ -0,0 +1,31 @@
+# Helm chart values for neon-proxy-scram.
+# This is a YAML-formatted file.
+
+image:
+  repository: neondatabase/neon
+
+settings:
+  authBackend: "console"
+  authEndpoint: "http://console-release.local/management/api/v2"
+  domain: "*.us-east-2.aws.neon.tech"
+
+# -- Additional labels for neon-proxy pods
+podLabels:
+  zenith_service: proxy-scram
+  zenith_env: prod
+  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.tech
+
+#metrics:
+#  enabled: true
+#  serviceMonitor:
+#    enabled: true
+#    selector:
+#      release: kube-prometheus-stack

.github/workflows/benchmarking.yml

@@ -265,7 +265,7 @@ jobs:
         PERF_TEST_RESULT_CONNSTR: "${{ secrets.PERF_TEST_RESULT_CONNSTR }}"
 
     - name: Create Allure report
-      if: always()
+      if: success() || failure()
       uses: ./.github/actions/allure-report
       with:
         action: generate

.github/workflows/build_and_test.yml

@@ -127,8 +127,8 @@ jobs:
             target/
           # Fall back to older versions of the key, if no cache for current Cargo.lock was found
           key: |
-            v9-${{ runner.os }}-${{ matrix.build_type }}-cargo-${{ hashFiles('Cargo.lock') }}
-            v9-${{ runner.os }}-${{ matrix.build_type }}-cargo-
+            v10-${{ runner.os }}-${{ matrix.build_type }}-cargo-${{ hashFiles('Cargo.lock') }}
+            v10-${{ runner.os }}-${{ matrix.build_type }}-cargo-
 
       - name: Cache postgres v14 build
         id: cache_pg_14
@@ -268,32 +268,6 @@ jobs:
         if: matrix.build_type == 'debug'
         uses: ./.github/actions/save-coverage-data
 
-  upload-latest-artifacts:
-    runs-on: dev
-    container:
-      image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned
-      options: --init
-    needs: [ regress-tests ]
-    if: github.ref_name == 'main'
-    steps:
-      - name: Copy Neon artifact to the latest directory
-        shell: bash -euxo pipefail {0}
-        env:
-          BUCKET: neon-github-public-dev
-          PREFIX: artifacts/${{ github.run_id }}
-        run: |
-          for build_type in debug release; do
-            FILENAME=neon-${{ runner.os }}-${build_type}-artifact.tar.zst
-
-            S3_KEY=$(aws s3api list-objects-v2 --bucket ${BUCKET} --prefix ${PREFIX} | jq -r '.Contents[].Key' | grep ${FILENAME} | sort --version-sort | tail -1 || true)
-            if [ -z "${S3_KEY}" ]; then
-              echo 2>&1 "Neither s3://${BUCKET}/${PREFIX}/${FILENAME} nor its version from previous attempts exist"
-              exit 1
-            fi
-
-            time aws s3 cp --only-show-errors s3://${BUCKET}/${S3_KEY} s3://${BUCKET}/artifacts/latest/${FILENAME}
-          done
-
   benchmarks:
     runs-on: dev
     container:
@@ -331,7 +305,7 @@ jobs:
       image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned
       options: --init
     needs: [ regress-tests, benchmarks ]
-    if: always()
+    if: success() || failure()
     strategy:
       fail-fast: false
       matrix:
@@ -389,7 +363,7 @@ jobs:
             !~/.cargo/registry/src
             ~/.cargo/git/
             target/
-          key: v9-${{ runner.os }}-${{ matrix.build_type }}-cargo-${{ hashFiles('Cargo.lock') }}
+          key: v10-${{ runner.os }}-${{ matrix.build_type }}-cargo-${{ hashFiles('Cargo.lock') }}
 
       - name: Get Neon artifact
         uses: ./.github/actions/download
@@ -445,11 +419,15 @@ jobs:
     container:
       image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/base:pinned
       options: --init
-    needs: [ build-neon ]
+    needs: [ push-docker-hub, tag ]
     steps:
       - name: Set PR's status to pending and request a remote CI test
         run: |
+          # For pull requests, GH Actions set "github.sha" variable to point at a fake merge commit
+          # but we need to use a real sha of a latest commit in the PR's branch for the e2e job,
+          # to place a job run status update later.
           COMMIT_SHA=${{ github.event.pull_request.head.sha }}
+          # For non-PR kinds of runs, the above will produce an empty variable, pick the original sha value for those
           COMMIT_SHA=${COMMIT_SHA:-${{ github.sha }}}
 
           REMOTE_REPO="${{ github.repository_owner }}/cloud"
@@ -475,12 +453,15 @@ jobs:
               \"inputs\": {
                 \"ci_job_name\": \"neon-cloud-e2e\",
                 \"commit_hash\": \"$COMMIT_SHA\",
-                \"remote_repo\": \"${{ github.repository }}\"
+                \"remote_repo\": \"${{ github.repository }}\",
+                \"storage_image_tag\": \"${{ needs.tag.outputs.build-tag }}\",
+                \"compute_image_tag\": \"${{ needs.tag.outputs.build-tag }}\"
               }
             }"
 
   neon-image:
     runs-on: dev
+    needs: [ tag ]
     container: gcr.io/kaniko-project/executor:v1.9.0-debug
 
     steps:
@@ -494,10 +475,11 @@ 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=${{ github.sha }} --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/neon:$GITHUB_RUN_ID
+        run: /kaniko/executor --snapshotMode=redo --cache=true --cache-repo 369495373322.dkr.ecr.eu-central-1.amazonaws.com/cache --snapshotMode=redo --context . --build-arg GIT_VERSION=${{ github.sha }} --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/neon:${{needs.tag.outputs.build-tag}}
 
   compute-tools-image:
     runs-on: dev
+    needs: [ tag ]
     container: gcr.io/kaniko-project/executor:v1.9.0-debug
 
     steps:
@@ -508,30 +490,12 @@ 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=${{ github.sha }} --dockerfile Dockerfile.compute-tools --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-tools:$GITHUB_RUN_ID
-
-  compute-node-image:
-    runs-on: dev
-    container: gcr.io/kaniko-project/executor:v1.9.0-debug
-    steps:
-      - name: Checkout
-        uses: actions/checkout@v1 # v3 won't work with kaniko
-        with:
-          submodules: true
-          fetch-depth: 0
-
-      - name: Configure ECR login
-        run: echo "{\"credsStore\":\"ecr-login\"}" > /kaniko/.docker/config.json
-
-        # compute-node uses postgres 14, which is default now
-        # cloud repo depends on this image name, thus duplicating it
-        # remove compute-node when cloud repo is updated
-      - name: Kaniko build compute node with extensions v14 (compatibility)
-        run: /kaniko/executor --skip-unused-stages --snapshotMode=redo --cache=true --cache-repo 369495373322.dkr.ecr.eu-central-1.amazonaws.com/cache --snapshotMode=redo --context . --build-arg GIT_VERSION=${{ github.sha }} --dockerfile Dockerfile.compute-node-v14 --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node:$GITHUB_RUN_ID
+        run: /kaniko/executor --snapshotMode=redo --cache=true --cache-repo 369495373322.dkr.ecr.eu-central-1.amazonaws.com/cache --snapshotMode=redo --context . --build-arg GIT_VERSION=${{ github.sha }} --dockerfile Dockerfile.compute-tools --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-tools:${{needs.tag.outputs.build-tag}}
 
   compute-node-image-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
@@ -543,12 +507,12 @@ 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=${{ github.sha }} --dockerfile Dockerfile.compute-node-v14 --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v14:$GITHUB_RUN_ID
-
+        run: /kaniko/executor --skip-unused-stages  --snapshotMode=redo --cache=true --cache-repo 369495373322.dkr.ecr.eu-central-1.amazonaws.com/cache  --context . --build-arg GIT_VERSION=${{ github.sha }} --dockerfile Dockerfile.compute-node-v14 --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v14:${{needs.tag.outputs.build-tag}}
 
   compute-node-image-v15:
     runs-on: dev
     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
@@ -560,25 +524,66 @@ 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=${{ github.sha }} --dockerfile Dockerfile.compute-node-v15 --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v15:$GITHUB_RUN_ID
+        run: /kaniko/executor --skip-unused-stages --snapshotMode=redo --cache=true --cache-repo 369495373322.dkr.ecr.eu-central-1.amazonaws.com/cache --context . --build-arg GIT_VERSION=${{ github.sha }} --dockerfile Dockerfile.compute-node-v15 --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v15:${{needs.tag.outputs.build-tag}}
+
+  test-images:
+    needs: [ tag, neon-image, compute-node-image-v14, compute-node-image-v15, compute-tools-image ]
+    runs-on: dev
+
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v3
+        with:
+          fetch-depth: 0
+
+      # `neondatabase/neon` contains multiple binaries, all of them use the same input for the version into the same version formatting library.
+      # Pick pageserver as currently the only binary with extra "version" features printed in the string to verify.
+      # Regular pageserver version string looks like
+      #   Neon page server git-env:32d14403bd6ab4f4520a94cbfd81a6acef7a526c failpoints: true, features: []
+      # Bad versions might loop like:
+      #   Neon page server git-env:local failpoints: true, features: ["testing"]
+      # Ensure that we don't have bad versions.
+      - name: Verify image versions
+        shell: bash # ensure no set -e for better error messages
+        run: |
+          pageserver_version=$(docker run --rm 369495373322.dkr.ecr.eu-central-1.amazonaws.com/neon:${{needs.tag.outputs.build-tag}} "/bin/sh" "-c" "/usr/local/bin/pageserver --version")
+
+          echo "Pageserver version string: $pageserver_version"
+
+          if ! echo "$pageserver_version" | grep -qv 'git-env:local' ; then
+            echo "Pageserver version should not be the default Dockerfile one"
+            exit 1
+          fi
+
+          if ! echo "$pageserver_version" | grep -qv '"testing"' ; then
+            echo "Pageserver version should have no testing feature enabled"
+            exit 1
+          fi
+
+      - name: Verify docker-compose example
+        run: env REPOSITORY=369495373322.dkr.ecr.eu-central-1.amazonaws.com TAG=${{needs.tag.outputs.build-tag}} ./docker-compose/docker_compose_test.sh
+
+      - name: Print logs and clean up
+        if: always()
+        run: |
+          docker compose -f ./docker-compose/docker-compose.yml logs || 0
+          docker compose -f ./docker-compose/docker-compose.yml down
 
   promote-images:
     runs-on: dev
-    needs: [ neon-image, compute-node-image, compute-node-image-v14, compute-node-image-v15, compute-tools-image ]
+    needs: [ tag, test-images ]
     if: github.event_name != 'workflow_dispatch'
     container: amazon/aws-cli
     strategy:
       fail-fast: false
       matrix:
-        # compute-node uses postgres 14, which is default now
-        # cloud repo depends on this image name, thus duplicating it
-        # remove compute-node when cloud repo is updated
-        name: [ neon, compute-node, compute-node-v14, compute-node-v15, compute-tools ]
+        name: [ neon, compute-node-v14, compute-node-v15, compute-tools ]
 
     steps:
       - name: Promote image to latest
-        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"
+        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"
 
   push-docker-hub:
     runs-on: dev
@@ -597,19 +602,16 @@ 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:latest neon
+        run: crane pull 369495373322.dkr.ecr.eu-central-1.amazonaws.com/neon:${{needs.tag.outputs.build-tag}} neon
 
       - name: Pull compute tools image from ECR
-        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:latest compute-node
+        run: crane pull 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-tools:${{needs.tag.outputs.build-tag}} compute-tools
 
       - name: Pull compute node v14 image from ECR
-        run: crane pull 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v14:latest compute-node-v14
+        run: crane pull 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v14:${{needs.tag.outputs.build-tag}} 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:latest compute-node-v15
+        run: crane pull 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v15:${{needs.tag.outputs.build-tag}} compute-node-v15
 
       - name: Pull rust image from ECR
         run: crane pull 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned rust
@@ -619,11 +621,10 @@ 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:$GITHUB_RUN_ID 093970136003.dkr.ecr.us-east-2.amazonaws.com/neon:latest
-          crane copy 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-tools:$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:$GITHUB_RUN_ID 093970136003.dkr.ecr.us-east-2.amazonaws.com/compute-node:latest
-          crane copy 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v14:$GITHUB_RUN_ID 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:$GITHUB_RUN_ID 093970136003.dkr.ecr.us-east-2.amazonaws.com/compute-node-v15:latest
+          crane copy 369495373322.dkr.ecr.eu-central-1.amazonaws.com/neon:${{needs.tag.outputs.build-tag}} 093970136003.dkr.ecr.eu-central-1.amazonaws.com/neon:latest
+          crane copy 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-tools:${{needs.tag.outputs.build-tag}} 093970136003.dkr.ecr.eu-central-1.amazonaws.com/compute-tools:latest
+          crane copy 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v14:${{needs.tag.outputs.build-tag}} 093970136003.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v14:latest
+          crane copy 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v15:${{needs.tag.outputs.build-tag}} 093970136003.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v15:latest
 
       - name: Configure Docker Hub login
         run: |
@@ -637,9 +638,6 @@ jobs:
       - name: Push compute tools image to Docker Hub
         run: crane push compute-tools neondatabase/compute-tools:${{needs.tag.outputs.build-tag}}
 
-      - name: Push compute node image to Docker Hub
-        run: crane push compute-node neondatabase/compute-node:${{needs.tag.outputs.build-tag}}
-
       - name: Push compute node v14 image to Docker Hub
         run: crane push compute-node-v14 neondatabase/compute-node-v14:${{needs.tag.outputs.build-tag}}
 
@@ -656,7 +654,6 @@ jobs:
         run: |
           crane tag neondatabase/neon:${{needs.tag.outputs.build-tag}} latest
           crane tag neondatabase/compute-tools:${{needs.tag.outputs.build-tag}} latest
-          crane tag neondatabase/compute-node:${{needs.tag.outputs.build-tag}} latest
           crane tag neondatabase/compute-node-v14:${{needs.tag.outputs.build-tag}} latest
           crane tag neondatabase/compute-node-v15:${{needs.tag.outputs.build-tag}} latest
 
@@ -671,11 +668,11 @@ jobs:
       - id: set-matrix
         run: |
           if [[ "$GITHUB_REF_NAME" == "main" ]]; then
-            STAGING='{"env_name": "staging", "proxy_job": "neon-proxy", "proxy_config": "staging.proxy", "kubeconfig_secret": "STAGING_KUBECONFIG_DATA"}'
-            NEON_STRESS='{"env_name": "neon-stress", "proxy_job": "neon-stress-proxy", "proxy_config": "neon-stress.proxy", "kubeconfig_secret": "NEON_STRESS_KUBECONFIG_DATA"}'
+            STAGING='{"env_name": "staging", "proxy_job": "neon-proxy", "proxy_config": "staging.proxy", "kubeconfig_secret": "STAGING_KUBECONFIG_DATA", "console_api_key_secret": "NEON_STAGING_API_KEY"}'
+            NEON_STRESS='{"env_name": "neon-stress", "proxy_job": "neon-stress-proxy", "proxy_config": "neon-stress.proxy", "kubeconfig_secret": "NEON_STRESS_KUBECONFIG_DATA", "console_api_key_secret": "NEON_CAPTEST_API_KEY"}'
             echo "include=[$STAGING, $NEON_STRESS]" >> $GITHUB_OUTPUT
           elif [[ "$GITHUB_REF_NAME" == "release" ]]; then
-            PRODUCTION='{"env_name": "production", "proxy_job": "neon-proxy", "proxy_config": "production.proxy", "kubeconfig_secret": "PRODUCTION_KUBECONFIG_DATA"}'
+            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]" >> $GITHUB_OUTPUT
           else
             echo "GITHUB_REF_NAME (value '$GITHUB_REF_NAME') is not set to either 'main' or 'release'"
@@ -735,7 +732,87 @@ jobs:
           ssh-add ssh-key
           rm -f ssh-key ssh-key-cert.pub
           ansible-galaxy collection install sivel.toiletwater
-          ansible-playbook deploy.yaml -i ${{ matrix.env_name }}.hosts.yaml
+          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
+    strategy:
+      matrix:
+        target_region: [ us-east-2 ]
+    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 '$GITHUB_REF_NAME') is not set to either 'main' or 'release'"
+            exit 1
+          fi
+
+          ansible-galaxy collection install sivel.toiletwater
+          ansible-playbook deploy.yaml -i staging.${{ matrix.target_region }}.hosts.yaml -e @ssm_config -e CONSOLE_API_TOKEN=${{secrets.NEON_STAGING_API_KEY}}
+          rm -f neon_install.tar.gz .neon_current_version
+
+  deploy-prod-new:
+    runs-on: prod
+    container: 093970136003.dkr.ecr.eu-central-1.amazonaws.com/ansible:latest
+    # We need both storage **and** compute images for deploy, because control plane picks the compute version based on the storage version.
+    # If it notices a fresh storage it may bump the compute version. And if compute image failed to build it may break things badly
+    needs: [ push-docker-hub, calculate-deploy-targets, tag, regress-tests ]
+    if: |
+      (github.ref_name == 'release') &&
+      github.event_name != 'workflow_dispatch'
+    defaults:
+      run:
+        shell: bash
+    strategy:
+      matrix:
+        target_region: [ us-east-2, eu-central-1, ap-southeast-1 ]
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v3
+        with:
+          submodules: true
+          fetch-depth: 0
+
+      - name: Redeploy
+        run: |
+          export DOCKER_TAG=${{needs.tag.outputs.build-tag}}
+          cd "$(pwd)/.github/ansible"
+
+          if [[ "$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 prod.${{ matrix.target_region }}.hosts.yaml -e @ssm_config -e CONSOLE_API_TOKEN=${{secrets.NEON_PRODUCTION_API_KEY}}
           rm -f neon_install.tar.gz .neon_current_version
 
   deploy-proxy:
@@ -779,3 +856,109 @@ 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
+    strategy:
+      matrix:
+        include:
+          - target_region:  us-east-2
+            target_cluster: dev-us-east-2-beta
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v3
+        with:
+          submodules: true
+          fetch-depth: 0
+
+      - name: Configure environment
+        run: |
+          helm repo add neondatabase https://neondatabase.github.io/helm-charts
+          aws --region ${{ matrix.target_region }} eks update-kubeconfig --name  ${{ matrix.target_cluster }}
+
+      - name: Re-deploy proxy
+        run: |
+          DOCKER_TAG=${{needs.tag.outputs.build-tag}}
+          helm upgrade neon-proxy-scram neondatabase/neon-proxy --namespace neon-proxy --create-namespace --install -f .github/helm-values/${{ matrix.target_cluster }}.neon-proxy-scram.yaml --set image.tag=${DOCKER_TAG} --wait --timeout 15m0s
+
+  deploy-proxy-prod-new:
+    runs-on: prod
+    container: 093970136003.dkr.ecr.eu-central-1.amazonaws.com/ansible:latest
+    # Compute image isn't strictly required for proxy deploy, but let's still wait for it to run all deploy jobs consistently.
+    needs: [ push-docker-hub, calculate-deploy-targets, tag, regress-tests ]
+    if: |
+      (github.ref_name == 'release') &&
+      github.event_name != 'workflow_dispatch'
+    defaults:
+      run:
+        shell: bash
+    strategy:
+      matrix:
+        include:
+          - target_region:  us-east-2
+            target_cluster: prod-us-east-2-delta
+          - target_region: eu-central-1
+            target_cluster: prod-eu-central-1-gamma
+          - target_region: ap-southeast-1
+            target_cluster: prod-ap-southeast-1-epsilon
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v3
+        with:
+          submodules: true
+          fetch-depth: 0
+
+      - name: Configure environment
+        run: |
+          helm repo add neondatabase https://neondatabase.github.io/helm-charts
+          aws --region ${{ matrix.target_region }} eks update-kubeconfig --name  ${{ matrix.target_cluster }}
+
+      - name: Re-deploy proxy
+        run: |
+          DOCKER_TAG=${{needs.tag.outputs.build-tag}}
+          helm upgrade neon-proxy-scram neondatabase/neon-proxy --namespace neon-proxy --create-namespace --install -f .github/helm-values/${{ matrix.target_cluster }}.neon-proxy-scram.yaml --set image.tag=${DOCKER_TAG} --wait --timeout 15m0s
+
+  promote-compatibility-data:
+    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: |
+          # Update compatibility snapshot for the release
+          for build_type in debug release; do
+            OLD_FILENAME=compatibility-snapshot-${build_type}-pg14-${GITHUB_RUN_ID}.tar.zst
+            NEW_FILENAME=compatibility-snapshot-${build_type}-pg14.tar.zst
+
+            time aws s3 mv --only-show-errors s3://${BUCKET}/${PREFIX}/${OLD_FILENAME} s3://${BUCKET}/${PREFIX}/${NEW_FILENAME}
+          done
+
+          # Update Neon artifact for the release (reuse already uploaded artifact)
+          for build_type in debug release; do
+            OLD_PREFIX=artifacts/${GITHUB_RUN_ID}
+            FILENAME=neon-${{ runner.os }}-${build_type}-artifact.tar.zst
+
+            S3_KEY=$(aws s3api list-objects-v2 --bucket ${BUCKET} --prefix ${OLD_PREFIX} | jq -r '.Contents[].Key' | grep ${FILENAME} | sort --version-sort | tail -1 || true)
+            if [ -z "${S3_KEY}" ]; then
+              echo 2>&1 "Neither s3://${BUCKET}/${OLD_PREFIX}/${FILENAME} nor its version from previous attempts exist"
+              exit 1
+            fi
+
+            time aws s3 cp --only-show-errors s3://${BUCKET}/${S3_KEY} s3://${BUCKET}/${PREFIX}/${FILENAME}
+          done

.github/workflows/codestyle.yml

@@ -36,7 +36,7 @@ jobs:
 
     steps:
       - name: Checkout
-        uses: actions/checkout@v2
+        uses: actions/checkout@v3
         with:
           submodules: true
           fetch-depth: 2
@@ -48,11 +48,11 @@ jobs:
         if: matrix.os == 'ubuntu-latest'
         run: |
           sudo apt update
-          sudo apt install build-essential libreadline-dev zlib1g-dev flex bison libseccomp-dev libssl-dev
+          sudo apt install build-essential libreadline-dev zlib1g-dev flex bison libseccomp-dev libssl-dev protobuf-compiler
 
       - name: Install macOS postgres dependencies
         if: matrix.os == 'macos-latest'
-        run: brew install flex bison openssl
+        run: brew install flex bison openssl protobuf
 
       - name: Set pg 14 revision for caching
         id: pg_v14_rev
@@ -106,7 +106,7 @@ jobs:
             !~/.cargo/registry/src
             ~/.cargo/git
             target
-          key: v5-${{ runner.os }}-cargo-${{ hashFiles('./Cargo.lock') }}-rust
+          key: v6-${{ runner.os }}-cargo-${{ hashFiles('./Cargo.lock') }}-rust
 
       - name: Run cargo clippy
         run: ./run_clippy.sh

.gitmodules

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

CODEOWNERS

@@ -0,0 +1,10 @@
+/compute_tools/ @neondatabase/control-plane
+/control_plane/ @neondatabase/compute @neondatabase/storage
+/libs/pageserver_api/ @neondatabase/compute @neondatabase/storage
+/libs/postgres_ffi/ @neondatabase/compute 
+/libs/remote_storage/ @neondatabase/storage 
+/libs/safekeeper_api/ @neondatabase/safekeepers  
+/pageserver/ @neondatabase/compute @neondatabase/storage 
+/pgxn/ @neondatabase/compute
+/proxy/ @neondatabase/control-plane 
+/safekeeper/ @neondatabase/safekeepers

Cargo.lock

@@ -40,8 +40,7 @@ dependencies = [
 [[package]]
 name = "amplify_num"
 version = "0.4.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f27d3d00d3d115395a7a8a4dc045feb7aa82b641e485f7e15f4e67ac16f4f56d"
+source = "git+https://github.com/hlinnaka/rust-amplify.git?branch=unsigned-int-perf#bd49b737c2e6e623ab8e9ba5ceaed5712d3a3940"
 
 [[package]]
 name = "android_system_properties"
@@ -75,9 +74,9 @@ checksum = "8da52d66c7071e2e3fa2a1e5c6d088fec47b593032b254f5e980de8ea54454d6"
 
 [[package]]
 name = "asn1-rs"
-version = "0.3.1"
+version = "0.5.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "30ff05a702273012438132f449575dbc804e27b2f3cbe3069aa237d26c98fa33"
+checksum = "cf6690c370453db30743b373a60ba498fc0d6d83b11f4abfd87a84a075db5dd4"
 dependencies = [
  "asn1-rs-derive",
  "asn1-rs-impl",
@@ -91,9 +90,9 @@ dependencies = [
 
 [[package]]
 name = "asn1-rs-derive"
-version = "0.1.0"
+version = "0.4.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "db8b7511298d5b7784b40b092d9e9dcd3a627a5707e4b5e507931ab0d44eeebf"
+checksum = "726535892e8eae7e70657b4c8ea93d26b8553afb1ce617caee529ef96d7dee6c"
 dependencies = [
  "proc-macro2",
  "quote",
@@ -262,15 +261,13 @@ dependencies = [
 
 [[package]]
 name = "bindgen"
-version = "0.60.1"
+version = "0.61.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "062dddbc1ba4aca46de6338e2bf87771414c335f7b2f2036e8f3e9befebf88e6"
+checksum = "8a022e58a142a46fea340d68012b9201c094e93ec3d033a944a24f8fd4a4f09a"
 dependencies = [
  "bitflags",
  "cexpr",
  "clang-sys",
- "clap",
- "env_logger",
  "lazy_static",
  "lazycell",
  "log",
@@ -280,6 +277,7 @@ dependencies = [
  "regex",
  "rustc-hash",
  "shlex",
+ "syn",
  "which",
 ]
 
@@ -319,21 +317,16 @@ dependencies = [
  "generic-array",
 ]
 
-[[package]]
-name = "boxfnonce"
-version = "0.1.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5988cb1d626264ac94100be357308f29ff7cbdd3b36bda27f450a4ee3f713426"
-
 [[package]]
 name = "bstr"
-version = "0.2.17"
+version = "1.0.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ba3569f383e8f1598449f1a423e72e99569137b47740b1da11ef19af3d5c3223"
+checksum = "fca0852af221f458706eb0725c03e4ed6c46af9ac98e6a689d5e634215d594dd"
 dependencies = [
- "lazy_static",
  "memchr",
+ "once_cell",
  "regex-automata",
+ "serde",
 ]
 
 [[package]]
@@ -450,22 +443,32 @@ version = "3.2.22"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "86447ad904c7fb335a790c9d7fe3d0d971dc523b8ccd1561a520de9a85302750"
 dependencies = [
- "atty",
  "bitflags",
- "clap_derive",
- "clap_lex",
+ "clap_lex 0.2.4",
  "indexmap",
- "once_cell",
- "strsim",
- "termcolor",
  "textwrap",
 ]
 
 [[package]]
-name = "clap_derive"
-version = "3.2.18"
+name = "clap"
+version = "4.0.15"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ea0c8bce528c4be4da13ea6fead8965e95b6073585a2f05204bd8f4119f82a65"
+checksum = "6bf8832993da70a4c6d13c581f4463c2bdda27b9bf1c5498dc4365543abe6d6f"
+dependencies = [
+ "atty",
+ "bitflags",
+ "clap_derive",
+ "clap_lex 0.3.0",
+ "once_cell",
+ "strsim",
+ "termcolor",
+]
+
+[[package]]
+name = "clap_derive"
+version = "4.0.13"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c42f169caba89a7d512b5418b09864543eeb4d497416c917d7137863bd2076ad"
 dependencies = [
  "heck",
  "proc-macro-error",
@@ -483,6 +486,15 @@ dependencies = [
  "os_str_bytes",
 ]
 
+[[package]]
+name = "clap_lex"
+version = "0.3.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0d4198f73e42b4936b35b5bb248d81d2b595ecb170da0bac7655c54eedfa8da8"
+dependencies = [
+ "os_str_bytes",
+]
+
 [[package]]
 name = "close_fds"
 version = "0.3.2"
@@ -540,7 +552,7 @@ version = "0.1.0"
 dependencies = [
  "anyhow",
  "chrono",
- "clap",
+ "clap 4.0.15",
  "env_logger",
  "futures",
  "hyper",
@@ -557,6 +569,42 @@ dependencies = [
  "workspace_hack",
 ]
 
+[[package]]
+name = "console-api"
+version = "0.4.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e57ff02e8ad8e06ab9731d5dc72dc23bef9200778eae1a89d555d8c42e5d4a86"
+dependencies = [
+ "prost 0.11.2",
+ "prost-types 0.11.2",
+ "tonic 0.8.2",
+ "tracing-core",
+]
+
+[[package]]
+name = "console-subscriber"
+version = "0.1.8"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "22a3a81dfaf6b66bce5d159eddae701e3a002f194d378cbf7be5f053c281d9be"
+dependencies = [
+ "console-api",
+ "crossbeam-channel",
+ "crossbeam-utils",
+ "futures",
+ "hdrhistogram",
+ "humantime",
+ "prost-types 0.11.2",
+ "serde",
+ "serde_json",
+ "thread_local",
+ "tokio",
+ "tokio-stream",
+ "tonic 0.8.2",
+ "tracing",
+ "tracing-core",
+ "tracing-subscriber",
+]
+
 [[package]]
 name = "const_format"
 version = "0.2.30"
@@ -582,7 +630,7 @@ name = "control_plane"
 version = "0.1.0"
 dependencies = [
  "anyhow",
- "clap",
+ "clap 4.0.15",
  "comfy-table",
  "git-version",
  "nix 0.25.0",
@@ -597,6 +645,7 @@ dependencies = [
  "tar",
  "thiserror",
  "toml",
+ "url",
  "utils",
  "workspace_hack",
 ]
@@ -675,7 +724,7 @@ dependencies = [
  "atty",
  "cast",
  "ciborium",
- "clap",
+ "clap 3.2.22",
  "criterion-plot",
  "itertools",
  "lazy_static",
@@ -743,7 +792,7 @@ dependencies = [
  "autocfg",
  "cfg-if",
  "crossbeam-utils",
- "memoffset",
+ "memoffset 0.6.5",
  "scopeguard",
 ]
 
@@ -846,16 +895,6 @@ dependencies = [
  "syn",
 ]
 
-[[package]]
-name = "daemonize"
-version = "0.4.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "70c24513e34f53b640819f0ac9f705b673fcf4006d7aab8778bee72ebfc89815"
-dependencies = [
- "boxfnonce",
- "libc",
-]
-
 [[package]]
 name = "darling"
 version = "0.14.1"
@@ -903,14 +942,14 @@ version = "0.7.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "d6ee87af31d84ef885378aebca32be3d682b0e0dc119d5b4860a2c5bb5046730"
 dependencies = [
- "uuid",
+ "uuid 0.8.2",
 ]
 
 [[package]]
 name = "der-parser"
-version = "7.0.0"
+version = "8.1.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "fe398ac75057914d7d07307bf67dc7f3f574a26783b4fc7805a20ffa9f506e82"
+checksum = "42d4bc9b0db0a0df9ae64634ac5bdefb7afcb534e182275ca0beadbe486701c1"
 dependencies = [
  "asn1-rs",
  "displaydoc",
@@ -1099,6 +1138,16 @@ version = "0.4.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "0ce7134b9999ecaf8bcd65542e436736ef32ddca1b3e06094cb6ec5755203b80"
 
+[[package]]
+name = "flate2"
+version = "1.0.24"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f82b0f4c27ad9f8bfd1f3208d882da2b09c301bc1c828fd3a00d0216d2fbbff6"
+dependencies = [
+ "crc32fast",
+ "miniz_oxide",
+]
+
 [[package]]
 name = "fnv"
 version = "1.0.7"
@@ -1219,6 +1268,12 @@ version = "0.3.24"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "a6508c467c73851293f390476d4491cf4d227dbabcd4170f3bb6044959b294f1"
 
+[[package]]
+name = "futures-timer"
+version = "3.0.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e64b03909df88034c26dc1547e8970b91f98bdb65165d6a4e9110d94263dbb2c"
+
 [[package]]
 name = "futures-util"
 version = "0.3.24"
@@ -1335,6 +1390,19 @@ dependencies = [
  "ahash",
 ]
 
+[[package]]
+name = "hdrhistogram"
+version = "7.5.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7f19b9f54f7c7f55e31401bb647626ce0cf0f67b0004982ce815b3ee72a02aa8"
+dependencies = [
+ "base64",
+ "byteorder",
+ "flate2",
+ "nom",
+ "num-traits",
+]
+
 [[package]]
 name = "heapless"
 version = "0.7.16"
@@ -1805,6 +1873,15 @@ dependencies = [
  "autocfg",
 ]
 
+[[package]]
+name = "memoffset"
+version = "0.7.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5de893c32cde5f383baa4c04c5d6dbdd735cfd4a794b0debdb2bb1b421da5ff4"
+dependencies = [
+ "autocfg",
+]
+
 [[package]]
 name = "metrics"
 version = "0.1.0"
@@ -1892,15 +1969,25 @@ name = "neon_broker"
 version = "0.1.0"
 dependencies = [
  "async-stream",
- "clap",
+ "bytes",
+ "clap 4.0.15",
+ "console-subscriber",
  "futures-core",
  "futures-util",
- "prost 0.11.0",
+ "git-version",
+ "humantime",
+ "hyper",
+ "metrics",
+ "once_cell",
+ "parking_lot 0.12.1",
+ "prost 0.11.2",
  "tokio",
  "tokio-stream",
  "tonic 0.8.2",
  "tonic-build 0.8.2",
+ "tracing",
  "utils",
+ "workspace_hack",
 ]
 
 [[package]]
@@ -1913,7 +2000,7 @@ dependencies = [
  "cc",
  "cfg-if",
  "libc",
- "memoffset",
+ "memoffset 0.6.5",
 ]
 
 [[package]]
@@ -1926,7 +2013,7 @@ dependencies = [
  "bitflags",
  "cfg-if",
  "libc",
- "memoffset",
+ "memoffset 0.6.5",
  "pin-utils",
 ]
 
@@ -2039,9 +2126,9 @@ dependencies = [
 
 [[package]]
 name = "oid-registry"
-version = "0.4.0"
+version = "0.6.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "38e20717fa0541f39bd146692035c37bedfa532b3e5071b35761082407546b2a"
+checksum = "7d4bda43fd1b844cbc6e6e54b5444e2b1bc7838bce59ad205902cccbb26d6761"
 dependencies = [
  "asn1-rs",
 ]
@@ -2132,12 +2219,12 @@ dependencies = [
  "byteorder",
  "bytes",
  "chrono",
- "clap",
+ "clap 4.0.15",
  "close_fds",
  "const_format",
  "crc32c",
+ "criterion",
  "crossbeam-utils",
- "daemonize",
  "etcd_broker",
  "fail",
  "futures",
@@ -2158,6 +2245,7 @@ dependencies = [
  "postgres-types",
  "postgres_ffi",
  "pprof",
+ "pq_proto",
  "rand",
  "regex",
  "remote_storage",
@@ -2167,8 +2255,10 @@ dependencies = [
  "serde_json",
  "serde_with",
  "signal-hook",
+ "svg_fmt",
  "tar",
  "tempfile",
+ "tenant_size_model",
  "thiserror",
  "tokio",
  "tokio-postgres",
@@ -2185,7 +2275,11 @@ dependencies = [
 name = "pageserver_api"
 version = "0.1.0"
 dependencies = [
+ "anyhow",
+ "byteorder",
+ "bytes",
  "const_format",
+ "postgres_ffi",
  "serde",
  "serde_with",
  "utils",
@@ -2408,7 +2502,7 @@ dependencies = [
  "env_logger",
  "hex",
  "log",
- "memoffset",
+ "memoffset 0.7.1",
  "once_cell",
  "postgres",
  "rand",
@@ -2445,6 +2539,21 @@ version = "0.2.16"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "eb9f9e6e233e5c4a35559a617bf40a4ec447db2e84c20b55a6f83167b7e57872"
 
+[[package]]
+name = "pq_proto"
+version = "0.1.0"
+dependencies = [
+ "anyhow",
+ "bytes",
+ "pin-project-lite",
+ "postgres-protocol",
+ "rand",
+ "serde",
+ "tokio",
+ "tracing",
+ "workspace_hack",
+]
+
 [[package]]
 name = "prettyplease"
 version = "0.1.21"
@@ -2535,12 +2644,12 @@ dependencies = [
 
 [[package]]
 name = "prost"
-version = "0.11.0"
+version = "0.11.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "399c3c31cdec40583bb68f0b18403400d01ec4289c383aa047560439952c4dd7"
+checksum = "a0841812012b2d4a6145fae9a6af1534873c32aa67fff26bd09f8fa42c83f95a"
 dependencies = [
  "bytes",
- "prost-derive 0.11.0",
+ "prost-derive 0.11.2",
 ]
 
 [[package]]
@@ -2567,9 +2676,9 @@ dependencies = [
 
 [[package]]
 name = "prost-build"
-version = "0.11.1"
+version = "0.11.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7f835c582e6bd972ba8347313300219fed5bfa52caf175298d860b61ff6069bb"
+checksum = "1d8b442418ea0822409d9e7d047cbf1e7e9e1760b172bf9982cf29d517c93511"
 dependencies = [
  "bytes",
  "heck",
@@ -2578,9 +2687,11 @@ dependencies = [
  "log",
  "multimap",
  "petgraph",
- "prost 0.11.0",
- "prost-types 0.11.1",
+ "prettyplease",
+ "prost 0.11.2",
+ "prost-types 0.11.2",
  "regex",
+ "syn",
  "tempfile",
  "which",
 ]
@@ -2600,9 +2711,9 @@ dependencies = [
 
 [[package]]
 name = "prost-derive"
-version = "0.11.0"
+version = "0.11.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7345d5f0e08c0536d7ac7229952590239e77abf0a0100a1b1d890add6ea96364"
+checksum = "164ae68b6587001ca506d3bf7f1000bfa248d0e1217b618108fba4ec1d0cc306"
 dependencies = [
  "anyhow",
  "itertools",
@@ -2623,12 +2734,12 @@ dependencies = [
 
 [[package]]
 name = "prost-types"
-version = "0.11.1"
+version = "0.11.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4dfaa718ad76a44b3415e6c4d53b17c8f99160dcb3a99b10470fce8ad43f6e3e"
+checksum = "747761bc3dc48f9a34553bf65605cf6cb6288ba219f3450b4275dbd81539551a"
 dependencies = [
  "bytes",
- "prost 0.11.0",
+ "prost 0.11.2",
 ]
 
 [[package]]
@@ -2641,7 +2752,7 @@ dependencies = [
  "base64",
  "bstr",
  "bytes",
- "clap",
+ "clap 4.0.15",
  "futures",
  "git-version",
  "hashbrown",
@@ -2654,6 +2765,7 @@ dependencies = [
  "once_cell",
  "parking_lot 0.12.1",
  "pin-project-lite",
+ "pq_proto",
  "rand",
  "rcgen",
  "reqwest",
@@ -2675,7 +2787,7 @@ dependencies = [
  "tracing-subscriber",
  "url",
  "utils",
- "uuid",
+ "uuid 1.2.1",
  "workspace_hack",
  "x509-parser",
 ]
@@ -2764,13 +2876,13 @@ dependencies = [
 
 [[package]]
 name = "rcgen"
-version = "0.8.14"
+version = "0.10.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5911d1403f4143c9d56a702069d593e8d0f3fab880a85e103604d0893ea31ba7"
+checksum = "ffbe84efe2f38dea12e9bfc1f65377fdf03e53a18cb3b995faedf7934c7e785b"
 dependencies = [
- "chrono",
  "pem",
  "ring",
+ "time 0.3.15",
  "yasna",
 ]
 
@@ -2960,9 +3072,21 @@ dependencies = [
 
 [[package]]
 name = "rstest"
-version = "0.12.0"
+version = "0.15.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d912f35156a3f99a66ee3e11ac2e0b3f34ac85a07e05263d05a7e2c8810d616f"
+checksum = "e9c9dc66cc29792b663ffb5269be669f1613664e69ad56441fdb895c2347b930"
+dependencies = [
+ "futures",
+ "futures-timer",
+ "rstest_macros",
+ "rustc_version 0.4.0",
+]
+
+[[package]]
+name = "rstest_macros"
+version = "0.14.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5015e68a0685a95ade3eee617ff7101ab6a3fc689203101ca16ebc16f2b89c66"
 dependencies = [
  "cfg-if",
  "proc-macro2",
@@ -3142,10 +3266,9 @@ dependencies = [
  "async-trait",
  "byteorder",
  "bytes",
- "clap",
+ "clap 4.0.15",
  "const_format",
  "crc32c",
- "daemonize",
  "etcd_broker",
  "fs2",
  "git-version",
@@ -3153,11 +3276,13 @@ dependencies = [
  "humantime",
  "hyper",
  "metrics",
+ "nix 0.25.0",
  "once_cell",
  "parking_lot 0.12.1",
  "postgres",
  "postgres-protocol",
  "postgres_ffi",
+ "pq_proto",
  "regex",
  "remote_storage",
  "safekeeper_api",
@@ -3523,6 +3648,12 @@ version = "2.4.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "6bdef32e8150c2a081110b42772ffe7d7c9032b606bc226c8260fd97e0976601"
 
+[[package]]
+name = "svg_fmt"
+version = "0.4.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8fb1df15f412ee2e9dfc1c504260fa695c1c3f10fe9f4a6ee2d2184d7d6450e2"
+
 [[package]]
 name = "symbolic-common"
 version = "8.8.0"
@@ -3532,7 +3663,7 @@ dependencies = [
  "debugid",
  "memmap2",
  "stable_deref_trait",
- "uuid",
+ "uuid 0.8.2",
 ]
 
 [[package]]
@@ -3600,6 +3731,13 @@ dependencies = [
  "winapi",
 ]
 
+[[package]]
+name = "tenant_size_model"
+version = "0.1.0"
+dependencies = [
+ "workspace_hack",
+]
+
 [[package]]
 name = "termcolor"
 version = "1.1.3"
@@ -3712,10 +3850,12 @@ dependencies = [
  "mio",
  "num_cpus",
  "once_cell",
+ "parking_lot 0.12.1",
  "pin-project-lite",
  "signal-hook-registry",
  "socket2",
  "tokio-macros",
+ "tracing",
  "winapi",
 ]
 
@@ -3895,8 +4035,8 @@ dependencies = [
  "hyper-timeout",
  "percent-encoding",
  "pin-project",
- "prost 0.11.0",
- "prost-derive 0.11.0",
+ "prost 0.11.2",
+ "prost-derive 0.11.2",
  "tokio",
  "tokio-stream",
  "tokio-util",
@@ -3928,7 +4068,7 @@ checksum = "48c6fd7c2581e36d63388a9e04c350c21beb7a8b059580b2e93993c526899ddc"
 dependencies = [
  "prettyplease",
  "proc-macro2",
- "prost-build 0.11.1",
+ "prost-build 0.11.2",
  "quote",
  "syn",
 ]
@@ -4039,6 +4179,16 @@ dependencies = [
  "tracing-core",
 ]
 
+[[package]]
+name = "tracing-serde"
+version = "0.1.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "bc6b213177105856957181934e4920de57730fc69bf42c37ee5bb664d406d9e1"
+dependencies = [
+ "serde",
+ "tracing-core",
+]
+
 [[package]]
 name = "tracing-subscriber"
 version = "0.3.16"
@@ -4049,12 +4199,15 @@ dependencies = [
  "nu-ansi-term",
  "once_cell",
  "regex",
+ "serde",
+ "serde_json",
  "sharded-slab",
  "smallvec",
  "thread_local",
  "tracing",
  "tracing-core",
  "tracing-log",
+ "tracing-serde",
 ]
 
 [[package]]
@@ -4137,9 +4290,7 @@ dependencies = [
  "metrics",
  "nix 0.25.0",
  "once_cell",
- "pin-project-lite",
- "postgres",
- "postgres-protocol",
+ "pq_proto",
  "rand",
  "routerify",
  "rustls",
@@ -4149,6 +4300,8 @@ dependencies = [
  "serde_json",
  "serde_with",
  "signal-hook",
+ "strum",
+ "strum_macros",
  "tempfile",
  "thiserror",
  "tokio",
@@ -4163,6 +4316,12 @@ name = "uuid"
 version = "0.8.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "bc5cf98d8186244414c848017f0e2676b3fcb46807f6668a97dfe67359a3c4b7"
+
+[[package]]
+name = "uuid"
+version = "1.2.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "feb41e78f93363bb2df8b0e86a2ca30eed7806ea16ea0c790d757cf93f79be83"
 dependencies = [
  "getrandom",
  "serde",
@@ -4212,7 +4371,7 @@ name = "wal_craft"
 version = "0.1.0"
 dependencies = [
  "anyhow",
- "clap",
+ "clap 4.0.15",
  "env_logger",
  "log",
  "once_cell",
@@ -4452,22 +4611,28 @@ dependencies = [
  "anyhow",
  "bytes",
  "chrono",
- "clap",
+ "clap 4.0.15",
  "crossbeam-utils",
  "either",
  "fail",
+ "futures-channel",
+ "futures-task",
+ "futures-util",
  "hashbrown",
  "indexmap",
  "libc",
  "log",
+ "memchr",
  "nom",
  "num-bigint",
  "num-integer",
  "num-traits",
  "prost 0.10.4",
+ "prost 0.11.2",
  "rand",
  "regex",
  "regex-syntax",
+ "reqwest",
  "scopeguard",
  "serde",
  "stable_deref_trait",
@@ -4477,14 +4642,13 @@ dependencies = [
  "tokio-util",
  "tracing",
  "tracing-core",
- "uuid",
 ]
 
 [[package]]
 name = "x509-parser"
-version = "0.13.2"
+version = "0.14.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9fb9bace5b5589ffead1afb76e43e34cff39cd0f3ce7e170ae0c29e53b88eb1c"
+checksum = "e0ecbeb7b67ce215e40e3cc7f2ff902f94a223acf44995934763467e7b1febc8"
 dependencies = [
  "asn1-rs",
  "base64",
@@ -4515,11 +4679,11 @@ checksum = "d2d7d3948613f75c98fd9328cfdcc45acc4d360655289d0a7d4ec931392200a3"
 
 [[package]]
 name = "yasna"
-version = "0.4.0"
+version = "0.5.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e262a29d0e61ccf2b6190d7050d4b237535fc76ce4c1210d9caa316f71dffa75"
+checksum = "346d34a236c9d3e5f3b9b74563f238f955bbd05fa0b8b4efa53c130c43982f4c"
 dependencies = [
- "chrono",
+ "time 0.3.15",
 ]
 
 [[package]]

Cargo.toml

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

Dockerfile

@@ -44,7 +44,7 @@ COPY . .
 # Show build caching stats to check if it was used in the end.
 # Has to be the part of the same RUN since cachepot daemon is killed in the end of this RUN, losing the compilation stats.
 RUN set -e \
-&& mold -run cargo build --bin pageserver --bin pageserver_binutils --bin safekeeper --bin proxy --locked --release \
+&& mold -run cargo build --bin pageserver --bin pageserver_binutils --bin draw_timeline_dir --bin safekeeper --bin neon_broker --bin proxy --locked --release \
     && cachepot -s
 
 # Build final image
@@ -65,7 +65,9 @@ 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/neon_broker         /usr/local/bin
 COPY --from=build --chown=neon:neon /home/nonroot/target/release/proxy               /usr/local/bin
 
 COPY --from=pg-build /home/nonroot/pg_install/v14 /usr/local/v14/

Dockerfile.compute-node-v14

@@ -1,50 +1,50 @@
-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
+#
+# This file is identical to the Dockerfile.compute-node-v15 file
+# except for the version of Postgres that is built.
+#
 
+ARG TAG=pinned
+
+#########################################################################################
 #
 # 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 libglib2.0-dev
+    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 libssl-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 && \
-    ./configure CFLAGS='-O2 -g3' --enable-debug --with-uuid=ossp && \
+    ./configure CFLAGS='-O2 -g3' --enable-debug --with-openssl --with-uuid=ossp && \
     make MAKELEVEL=0 -j $(getconf _NPROCESSORS_ONLN) -s install && \
     make MAKELEVEL=0 -j $(getconf _NPROCESSORS_ONLN) -s -C contrib/ install && \
     # Install headers
     make MAKELEVEL=0 -j $(getconf _NPROCESSORS_ONLN) -s -C src/include install && \
     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-3.3.0.tar.gz && \
-    tar xvzf postgis-3.3.0.tar.gz && \
-    cd postgis-3.3.0 && \
+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 && \
     ./autogen.sh && \
     export PATH="/usr/local/pgsql/bin:$PATH" && \
     ./configure && \
@@ -57,39 +57,55 @@ RUN wget https://download.osgeo.org/postgis/source/postgis-3.3.0.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
+    apt install -y ninja-build python3-dev libc++-dev libc++abi-dev libncurses5 binutils
 
-# 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
+# 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
 
+# 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" && \
-    make -j $(getconf _NPROCESSORS_ONLN) && \
-    make -j $(getconf _NPROCESSORS_ONLN) install && \
+    sed -i 's/MemoryContextAlloc(/MemoryContextAllocZero(/' plv8.cc && \
+    make DOCKER=1 -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 apt update && \
-    apt install -y --no-install-recommends -t testing cmake
+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 wget https://github.com/uber/h3/archive/refs/tags/v4.0.1.tar.gz -O h3.tgz && \
     tar xvzf h3.tgz  && \
@@ -108,16 +124,18 @@ 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.control && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/h3_postgis.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/
-# plv8 still sometimes crashes during the creation
-# COPY --from=plv8-build /usr/local/pgsql/ /usr/local/pgsql/
+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/
@@ -127,16 +145,22 @@ 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
 
@@ -154,10 +178,12 @@ 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 && \
@@ -174,8 +200,6 @@ COPY --from=compute-tools --chown=postgres /home/nonroot/target/release-line-deb
 # libreadline8 for psql
 # libossp-uuid16 for extension ossp-uuid
 # libgeos, libgdal, libproj and libprotobuf-c1 for PostGIS
-# 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.
@@ -188,12 +212,6 @@ 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

@@ -4,44 +4,39 @@
 #
 
 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 libglib2.0-dev
+    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 libssl-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 && \
-    ./configure CFLAGS='-O2 -g3' --enable-debug --with-uuid=ossp && \
+    ./configure CFLAGS='-O2 -g3' --enable-debug --with-openssl --with-uuid=ossp && \
     make MAKELEVEL=0 -j $(getconf _NPROCESSORS_ONLN) -s install && \
     make MAKELEVEL=0 -j $(getconf _NPROCESSORS_ONLN) -s -C contrib/ install && \
     # Install headers
     make MAKELEVEL=0 -j $(getconf _NPROCESSORS_ONLN) -s -C src/include install && \
     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 && \
@@ -62,39 +57,55 @@ 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
+    apt install -y ninja-build python3-dev libc++-dev libc++abi-dev libncurses5 binutils
 
-# 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
+# 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
 
+# 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" && \
-    make -j $(getconf _NPROCESSORS_ONLN) && \
-    make -j $(getconf _NPROCESSORS_ONLN) install && \
+    sed -i 's/MemoryContextAlloc(/MemoryContextAllocZero(/' plv8.cc && \
+    make DOCKER=1 -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 apt update && \
-    apt install -y --no-install-recommends -t testing cmake
+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 wget https://github.com/uber/h3/archive/refs/tags/v4.0.1.tar.gz -O h3.tgz && \
     tar xvzf h3.tgz  && \
@@ -113,16 +124,18 @@ 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.control && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/h3_postgis.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/
-# plv8 still sometimes crashes during the creation
-# COPY --from=plv8-build /usr/local/pgsql/ /usr/local/pgsql/
+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/
@@ -132,16 +145,22 @@ 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
 
@@ -159,10 +178,12 @@ 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 && \
@@ -179,8 +200,6 @@ COPY --from=compute-tools --chown=postgres /home/nonroot/target/release-line-deb
 # libreadline8 for psql
 # libossp-uuid16 for extension ossp-uuid
 # libgeos, libgdal, libproj and libprotobuf-c1 for PostGIS
-# 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.
@@ -193,12 +212,6 @@ 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.legacy

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

Makefile

@@ -151,6 +151,11 @@ neon-pg-ext-v14: postgres-v14
 	(cd $(POSTGRES_INSTALL_DIR)/build/neon-v14 && \
 	$(MAKE) PG_CONFIG=$(POSTGRES_INSTALL_DIR)/v14/bin/pg_config CFLAGS='$(PG_CFLAGS) $(COPT)' \
 		-f $(ROOT_PROJECT_DIR)/pgxn/neon/Makefile install)
+	+@echo "Compiling neon_walredo v14"
+	mkdir -p $(POSTGRES_INSTALL_DIR)/build/neon-walredo-v14
+	(cd $(POSTGRES_INSTALL_DIR)/build/neon-walredo-v14 && \
+	$(MAKE) PG_CONFIG=$(POSTGRES_INSTALL_DIR)/v14/bin/pg_config CFLAGS='$(PG_CFLAGS) $(COPT)' \
+		-f $(ROOT_PROJECT_DIR)/pgxn/neon_walredo/Makefile install)
 	+@echo "Compiling neon_test_utils" v14
 	mkdir -p $(POSTGRES_INSTALL_DIR)/build/neon-test-utils-v14
 	(cd $(POSTGRES_INSTALL_DIR)/build/neon-test-utils-v14 && \
@@ -163,6 +168,11 @@ neon-pg-ext-v15: postgres-v15
 	(cd $(POSTGRES_INSTALL_DIR)/build/neon-v15 && \
 	$(MAKE) PG_CONFIG=$(POSTGRES_INSTALL_DIR)/v15/bin/pg_config CFLAGS='$(PG_CFLAGS) $(COPT)' \
 		-f $(ROOT_PROJECT_DIR)/pgxn/neon/Makefile install)
+	+@echo "Compiling neon_walredo v15"
+	mkdir -p $(POSTGRES_INSTALL_DIR)/build/neon-walredo-v15
+	(cd $(POSTGRES_INSTALL_DIR)/build/neon-walredo-v15 && \
+	$(MAKE) PG_CONFIG=$(POSTGRES_INSTALL_DIR)/v15/bin/pg_config CFLAGS='$(PG_CFLAGS) $(COPT)' \
+		-f $(ROOT_PROJECT_DIR)/pgxn/neon_walredo/Makefile install)
 	+@echo "Compiling neon_test_utils" v15
 	mkdir -p $(POSTGRES_INSTALL_DIR)/build/neon-test-utils-v15
 	(cd $(POSTGRES_INSTALL_DIR)/build/neon-test-utils-v15 && \

README.md

@@ -35,12 +35,12 @@ Pageserver consists of:
 * On Ubuntu or Debian, this set of packages should be sufficient to build the code:
 ```bash
 apt install build-essential libtool libreadline-dev zlib1g-dev flex bison libseccomp-dev \
-libssl-dev clang pkg-config libpq-dev etcd cmake postgresql-client
+libssl-dev clang pkg-config libpq-dev etcd cmake postgresql-client protobuf-compiler
 ```
 * On Fedora, these packages are needed:
 ```bash
 dnf install flex bison readline-devel zlib-devel openssl-devel \
-  libseccomp-devel perl clang cmake etcd postgresql postgresql-contrib
+  libseccomp-devel perl clang cmake etcd postgresql postgresql-contrib protobuf-compiler
 ```
 
 2. [Install Rust](https://www.rust-lang.org/tools/install)
@@ -223,10 +223,7 @@ Ensure your dependencies are installed as described [here](https://github.com/ne
 ```sh
 git clone --recursive https://github.com/neondatabase/neon.git
 
-# either:
 CARGO_BUILD_FLAGS="--features=testing" make
-# or:
-make debug
 
 ./scripts/pytest
 ```

broker/Cargo.toml

@@ -5,31 +5,38 @@ edition = "2021"
 
 [features]
 bench = []
-
-[[bin]]
-name = "neon_broker"
-path = "src/broker.rs"
+# for exploring with tokio-console. Note that tokio_unstable cfg must be enabled, i.e.
+# RUSTFLAGS="--cfg tokio_unstable" cargo build -r -p neon_broker --features console
+console = ["dep:console-subscriber", "tokio/full", "tokio/tracing"]
 
 [[bin]]
 name = "neon_broker_bench"
-path = "src/bench.rs"
-# build benchmarking binary only if explicitly requested with '--feature bench'
-# required-features = ["bench"]
+path = "src/bin/bench.rs"
+# build benchmarking binary only if explicitly requested with '--features bench'
+required-features = ["bench"]
 
 [dependencies]
 async-stream = "0.3"
+bytes = "1.0"
+clap = { version = "4.0", features = ["derive"] }
 futures-core = "0.3"
 futures-util = "0.3"
-tonic = "0.8"
+git-version = "0.3.5"
+humantime = "2.1.0"
+hyper = {version = "0.14.14", features = ["full"]}
+once_cell = "1.13.0"
+parking_lot = "0.12"
 prost = "0.11"
+tonic = "0.8"
 tokio = { version = "1.0", features = ["macros", "rt-multi-thread"] }
-# for exploring with tokio-console
-# tokio = { version = "1", features = ["full", "tracing"] }
-# console-subscriber = "0.1.8"
 tokio-stream = "0.1"
-clap = { version = "3.2.17", features = ["derive"] }
+tracing = "0.1.27"
 
+console-subscriber = {version = "0.1.8", optional = true }
+
+metrics = { path = "../libs/metrics" }
 utils = { path = "../libs/utils" }
+workspace_hack = { version = "0.1", path = "../workspace_hack" }
 
 [build-dependencies]
 tonic-build = "0.8"

broker/build.rs

@@ -1,4 +1,7 @@
 fn main() -> Result<(), Box<dyn std::error::Error>> {
-    tonic_build::compile_protos("proto/broker.proto")?;
+    // Generate code to deterministic location to make finding it easier.
+    tonic_build::configure()
+        .out_dir("proto/") // put generated code to proto/
+        .compile(&["proto/broker.proto"], &["proto/"])?;
     Ok(())
 }

broker/proto/.gitignore

@@ -0,0 +1,2 @@
+# protobuf generated code
+neon_broker.rs

broker/proto/broker.proto

@@ -1,5 +1,7 @@
 syntax = "proto3";
 
+import "google/protobuf/empty.proto";
+
 package neon_broker;
 
 service NeonBroker {
@@ -7,12 +9,12 @@ service NeonBroker {
     rpc SubscribeSafekeeperInfo(SubscribeSafekeeperInfoRequest) returns (stream SafekeeperTimelineInfo) {};
 
     // Publish safekeeper updates.
-    rpc PublishSafekeeperInfo(stream SafekeeperTimelineInfo) returns (Empty) {};
+    rpc PublishSafekeeperInfo(stream SafekeeperTimelineInfo) returns (google.protobuf.Empty) {};
 }
 
 message SubscribeSafekeeperInfoRequest {
     oneof subscription_key {
-        Empty all = 1; // subscribe to everything
+        google.protobuf.Empty all = 1; // subscribe to everything
         TenantTimelineId tenant_timeline_id = 2; // subscribe to specific timeline
     }
 }
@@ -20,19 +22,23 @@ message SubscribeSafekeeperInfoRequest {
 message SafekeeperTimelineInfo {
     uint64 safekeeper_id = 1;
     TenantTimelineId tenant_timeline_id = 2;
+    // Term of the last entry.
     uint64 last_log_term = 3;
+    // LSN of the last record.
     uint64 flush_lsn = 4;
+    // Up to which LSN safekeeper regards its WAL as committed.
     uint64 commit_lsn = 5;
+    // LSN up to which safekeeper has backed WAL.
     uint64 backup_lsn = 6;
+    // LSN of last checkpoint uploaded by pageserver.
     uint64 remote_consistent_lsn = 7;
     uint64 peer_horizon_lsn = 8;
-    string safekeeper_connstr = 9;
+    uint64 local_start_lsn = 9;
+    // A connection string to use for WAL receiving.
+    string safekeeper_connstr = 10;
 }
 
 message TenantTimelineId {
     bytes tenant_id = 1;
     bytes timeline_id = 2;
-}
-
-message Empty {
 }

broker/readme.md

@@ -1,4 +0,0 @@
-```
-cargo build -r -p neon_broker --features bench && target/release/neon_broker
-target/release/neon_broker_bench -s 1 -p 1
-```

broker/src/bin/bench.rs

@@ -1,23 +1,31 @@
-pub mod neon_broker {
-    tonic::include_proto!("neon_broker");
-}
-
 use std::sync::atomic::{AtomicU64, Ordering};
 use std::sync::Arc;
 use std::time::{Duration, Instant};
 
 use clap::Parser;
-use neon_broker::neon_broker_client::NeonBrokerClient;
-use neon_broker::subscribe_safekeeper_info_request::SubscriptionKey;
-use neon_broker::TenantTimelineId as ProtoTenantTimelineId;
-use neon_broker::{SafekeeperTimelineInfo, SubscribeSafekeeperInfoRequest};
-use tokio::time::{self, sleep};
+use neon_broker::neon_broker_proto::subscribe_safekeeper_info_request::SubscriptionKey;
+use neon_broker::neon_broker_proto::TenantTimelineId as ProtoTenantTimelineId;
+use neon_broker::neon_broker_proto::{SafekeeperTimelineInfo, SubscribeSafekeeperInfoRequest};
+use neon_broker::NeonBrokerClientChannel;
+use neon_broker::DEFAULT_LISTEN_ADDR;
+use tokio::time::{self};
 
-use tonic::transport::Channel;
 use tonic::Request;
 
+const ABOUT: &str = r#"
+A simple benchmarking tool for neon_broker. Creates specified number of per
+timeline publishers and subscribers; each publisher continiously sends
+messages, subscribers read them. Each second the tool outputs number of
+messages summed across all subscribers and min number of messages
+recevied by single subscriber.
+
+For example,
+cargo build -r -p neon_broker --features bench && target/release/neon_broker
+target/release/neon_broker_bench -s 1 -p 1
+"#;
+
 #[derive(Parser, Debug)]
-#[clap(author, version, about, long_about = None)]
+#[clap(author, version, about = ABOUT)]
 struct Args {
     /// Number of publishers
     #[clap(short = 'p', long, value_parser, default_value_t = 1)]
@@ -39,13 +47,6 @@ async fn progress_reporter(counters: Vec<Arc<AtomicU64>>) {
     let mut skipped: u64 = 0;
     loop {
         interval.tick().await;
-        // print!(
-        //     "cnts are {:?}",
-        //     counters
-        //         .iter()
-        //         .map(|c| c.load(Ordering::Relaxed))
-        //         .collect::<Vec<_>>()
-        // );
         let c_new = counters.iter().map(|c| c.load(Ordering::Relaxed)).sum();
         let c_min_new = counters
             .iter()
@@ -81,15 +82,15 @@ fn tli_from_u64(i: u64) -> Vec<u8> {
     timeline_id
 }
 
-async fn subscribe(client: Option<NeonBrokerClient<Channel>>, counter: Arc<AtomicU64>, i: u64) {
+async fn subscribe(client: Option<NeonBrokerClientChannel>, counter: Arc<AtomicU64>, i: u64) {
     let mut client = match client {
         Some(c) => c,
-        None => NeonBrokerClient::connect("http://[::1]:50051")
+        None => NeonBrokerClientChannel::connect_lazy(format!("http://{}", DEFAULT_LISTEN_ADDR))
             .await
             .unwrap(),
     };
 
-    // let key = SubscriptionKey::All(Empty {});
+    // let key = SubscriptionKey::All(());
     let key = SubscriptionKey::TenantTimelineId(ProtoTenantTimelineId {
         tenant_id: vec![0xFF; 16],
         timeline_id: tli_from_u64(i),
@@ -105,14 +106,13 @@ async fn subscribe(client: Option<NeonBrokerClient<Channel>>, counter: Arc<Atomi
 
     while let Some(_feature) = stream.message().await.unwrap() {
         counter.fetch_add(1, Ordering::Relaxed);
-        // println!("info = {:?}, client {}", _feature, i);
     }
 }
 
-async fn publish(client: Option<NeonBrokerClient<Channel>>, n_keys: u64) {
+async fn publish(client: Option<NeonBrokerClientChannel>, n_keys: u64) {
     let mut client = match client {
         Some(c) => c,
-        None => NeonBrokerClient::connect("http://[::1]:50051")
+        None => NeonBrokerClientChannel::connect_lazy(format!("http://{}", DEFAULT_LISTEN_ADDR))
             .await
             .unwrap(),
     };
@@ -134,20 +134,14 @@ async fn publish(client: Option<NeonBrokerClient<Channel>>, n_keys: u64) {
                 remote_consistent_lsn: 4,
                 peer_horizon_lsn: 5,
                 safekeeper_connstr: "zenith-1-sk-1.local:7676".to_owned(),
+                local_start_lsn: 0,
             };
             counter += 1;
-            // println!("sending info = {:?}", info);
-            // if counter >= 1000 {
-                // break;
-            // }
             yield info;
-            // sleep(Duration::from_millis(100)).await;
         }
     };
-    let _response = client
-        .publish_safekeeper_info(Request::new(outbound))
-        .await
-        .unwrap();
+    let response = client.publish_safekeeper_info(Request::new(outbound)).await;
+    println!("pub response is {:?}", response);
 }
 
 #[tokio::main]
@@ -160,7 +154,7 @@ async fn main() -> Result<(), Box<dyn std::error::Error>> {
     }
     let h = tokio::spawn(progress_reporter(counters.clone()));
 
-    let c = NeonBrokerClient::connect("http://[::1]:50051")
+    let c = NeonBrokerClientChannel::connect_lazy(format!("http://{}", DEFAULT_LISTEN_ADDR))
         .await
         .unwrap();
 
@@ -171,7 +165,8 @@ async fn main() -> Result<(), Box<dyn std::error::Error>> {
     }
     for _i in 0..args.num_pubs {
         // let c = Some(c.clone());
-        tokio::spawn(publish(None, args.num_subs as u64));
+        let c = None;
+        tokio::spawn(publish(c, args.num_subs as u64));
     }
 
     h.await?;

broker/src/bin/neon_broker.rs

@@ -0,0 +1,510 @@
+//! Simple pub-sub based on grpc (tonic) and Tokio mpsc for storage nodes
+//! messaging. The main design goal is to avoid central synchronization during
+//! normal flow, resorting to it only when pub/sub change happens. Each
+//! subscriber holds mpsc for messages it sits on; tx end is sent to existing
+//! publishers and saved in shared state for new ones. Publishers maintain
+//! locally set of subscribers they stream messages to.
+//!
+//! Subscriptions to 1) single timeline 2) everything are possible. We could add
+//! subscription to set of timelines to save grpc streams, but testing shows
+//! many individual streams is also ok.
+//!
+//! Message is dropped if subscriber can't consume it, not affecting other
+//! subscribers.
+//!
+//! Only safekeeper message is supported, but it is not hard to add something
+//! else with generics.
+use clap::{command, Parser};
+use futures_core::Stream;
+use futures_util::StreamExt;
+use hyper::header::CONTENT_TYPE;
+use hyper::service::{make_service_fn, service_fn};
+use hyper::{Body, Method, StatusCode};
+use parking_lot::RwLock;
+use std::collections::HashMap;
+use std::convert::Infallible;
+use std::net::SocketAddr;
+use std::pin::Pin;
+use std::sync::Arc;
+use std::time::Duration;
+use tokio::sync::broadcast;
+use tokio::sync::broadcast::error::RecvError;
+use tonic::codegen::Service;
+use tonic::Code;
+use tonic::{Request, Response, Status};
+use tracing::*;
+
+use metrics::{Encoder, TextEncoder};
+use neon_broker::metrics::{NUM_PUBS, NUM_SUBS_ALL, NUM_SUBS_TIMELINE};
+use neon_broker::neon_broker_proto::neon_broker_server::{NeonBroker, NeonBrokerServer};
+use neon_broker::neon_broker_proto::subscribe_safekeeper_info_request::SubscriptionKey as ProtoSubscriptionKey;
+use neon_broker::neon_broker_proto::{SafekeeperTimelineInfo, SubscribeSafekeeperInfoRequest};
+use neon_broker::{parse_proto_ttid, EitherBody, DEFAULT_LISTEN_ADDR};
+use utils::id::TenantTimelineId;
+use utils::project_git_version;
+
+project_git_version!(GIT_VERSION);
+
+const DEFAULT_CHAN_SIZE: usize = 256;
+const DEFAULT_HTTP2_KEEPALIVE_INTERVAL: &str = "5000ms";
+
+#[derive(Parser, Debug)]
+#[command(version = GIT_VERSION, about = "Broker for neon storage nodes communication", long_about = None)]
+struct Args {
+    /// Endpoint to listen on.
+    #[arg(short, long, default_value = DEFAULT_LISTEN_ADDR)]
+    listen_addr: SocketAddr,
+    /// Size of the queue to the subscriber.
+    #[arg(long, default_value_t = DEFAULT_CHAN_SIZE)]
+    chan_size: usize,
+    /// HTTP/2 keepalive interval.
+    #[arg(long, value_parser= humantime::parse_duration, default_value = DEFAULT_HTTP2_KEEPALIVE_INTERVAL)]
+    http2_keepalive_interval: Duration,
+    /// Format for logging, either 'plain' or 'json'.
+    #[arg(long, default_value = "plain")]
+    log_format: String,
+}
+
+type PubId = u64; // id of publisher for registering in maps
+type SubId = u64; // id of subscriber for registering in maps
+
+#[derive(Copy, Clone)]
+enum SubscriptionKey {
+    All,
+    Timeline(TenantTimelineId),
+}
+
+impl SubscriptionKey {
+    // Parse protobuf subkey (protobuf doesn't have fixed size bytes, we get vectors).
+    pub fn from_proto_subscription_key(key: ProtoSubscriptionKey) -> Result<Self, Status> {
+        match key {
+            ProtoSubscriptionKey::All(_) => Ok(SubscriptionKey::All),
+            ProtoSubscriptionKey::TenantTimelineId(proto_ttid) => {
+                Ok(SubscriptionKey::Timeline(parse_proto_ttid(&proto_ttid)?))
+            }
+        }
+    }
+}
+
+// Channel to timeline subscribers.
+struct ChanToTimelineSub {
+    chan: broadcast::Sender<SafekeeperTimelineInfo>,
+    // Tracked separately to know when delete the shmem entry. receiver_count()
+    // is unhandy for that as unregistering and dropping the receiver side
+    // happens at different moments.
+    num_subscribers: u64,
+}
+
+struct SharedState {
+    next_pub_id: PubId,
+    num_pubs: i64,
+    next_sub_id: SubId,
+    num_subs_to_timelines: i64,
+    chans_to_timeline_subs: HashMap<TenantTimelineId, ChanToTimelineSub>,
+    num_subs_to_all: i64,
+    chan_to_all_subs: broadcast::Sender<SafekeeperTimelineInfo>,
+}
+
+impl SharedState {
+    pub fn new(chan_size: usize) -> Self {
+        SharedState {
+            next_pub_id: 0,
+            num_pubs: 0,
+            next_sub_id: 0,
+            num_subs_to_timelines: 0,
+            chans_to_timeline_subs: HashMap::new(),
+            num_subs_to_all: 0,
+            chan_to_all_subs: broadcast::channel(chan_size).0,
+        }
+    }
+
+    // Register new publisher.
+    pub fn register_publisher(&mut self, registry: Registry) -> Publisher {
+        let pub_id = self.next_pub_id;
+        self.next_pub_id += 1;
+        self.num_pubs += 1;
+        NUM_PUBS.set(self.num_pubs);
+        Publisher {
+            id: pub_id,
+            registry,
+        }
+    }
+
+    // Unregister publisher.
+    pub fn unregister_publisher(&mut self) {
+        self.num_pubs -= 1;
+        NUM_PUBS.set(self.num_pubs);
+    }
+
+    // Register new subscriber.
+    pub fn register_subscriber(
+        &mut self,
+        sub_key: SubscriptionKey,
+        registry: Registry,
+        chan_size: usize,
+    ) -> Subscriber {
+        let sub_id = self.next_sub_id;
+        self.next_sub_id += 1;
+        let sub_rx = match sub_key {
+            SubscriptionKey::All => {
+                self.num_subs_to_all += 1;
+                NUM_SUBS_ALL.set(self.num_subs_to_all);
+                self.chan_to_all_subs.subscribe()
+            }
+            SubscriptionKey::Timeline(ttid) => {
+                self.num_subs_to_timelines += 1;
+                NUM_SUBS_TIMELINE.set(self.num_subs_to_timelines);
+                // Create new broadcast channel for this key, or subscriber to
+                // the existing one.
+                let chan_to_timeline_sub =
+                    self.chans_to_timeline_subs
+                        .entry(ttid)
+                        .or_insert(ChanToTimelineSub {
+                            chan: broadcast::channel(chan_size).0,
+                            num_subscribers: 0,
+                        });
+                chan_to_timeline_sub.num_subscribers += 1;
+                chan_to_timeline_sub.chan.subscribe()
+            }
+        };
+        Subscriber {
+            id: sub_id,
+            key: sub_key,
+            sub_rx,
+            registry,
+        }
+    }
+
+    // Unregister the subscriber.
+    pub fn unregister_subscriber(&mut self, sub_key: SubscriptionKey) {
+        match sub_key {
+            SubscriptionKey::All => {
+                self.num_subs_to_all -= 1;
+                NUM_SUBS_ALL.set(self.num_subs_to_all);
+            }
+            SubscriptionKey::Timeline(ttid) => {
+                self.num_subs_to_timelines -= 1;
+                NUM_SUBS_TIMELINE.set(self.num_subs_to_timelines);
+
+                // Remove from the map, destroying the channel, if we are the
+                // last subscriber to this timeline.
+
+                // Missing entry is a bug; we must have registered.
+                let chan_to_timeline_sub = self.chans_to_timeline_subs.get_mut(&ttid).unwrap();
+                chan_to_timeline_sub.num_subscribers -= 1;
+                if chan_to_timeline_sub.num_subscribers == 0 {
+                    self.chans_to_timeline_subs.remove(&ttid);
+                }
+            }
+        }
+    }
+}
+
+// SharedState wrapper.
+#[derive(Clone)]
+struct Registry {
+    shared_state: Arc<RwLock<SharedState>>,
+    chan_size: usize,
+}
+
+impl Registry {
+    // Register new publisher in shared state.
+    pub fn register_publisher(&self) -> Publisher {
+        let publisher = self.shared_state.write().register_publisher(self.clone());
+        trace!("registered publisher {}", publisher.id);
+        publisher
+    }
+
+    pub fn unregister_publisher(&self, publisher: &Publisher) {
+        self.shared_state.write().unregister_publisher();
+        trace!("unregistered publisher {}", publisher.id);
+    }
+
+    // Register new subscriber in shared state.
+    pub fn register_subscriber(&self, sub_key: SubscriptionKey) -> Subscriber {
+        let subscriber =
+            self.shared_state
+                .write()
+                .register_subscriber(sub_key, self.clone(), self.chan_size);
+        trace!("registered subscriber {}", subscriber.id);
+        subscriber
+    }
+
+    // Unregister the subscriber
+    pub fn unregister_subscriber(&self, subscriber: &Subscriber) {
+        self.shared_state
+            .write()
+            .unregister_subscriber(subscriber.key);
+        trace!("unregistered subscriber {}", subscriber.id);
+    }
+}
+
+// Private subscriber state.
+struct Subscriber {
+    id: SubId,
+    key: SubscriptionKey,
+    // Subscriber receives messages from publishers here.
+    sub_rx: broadcast::Receiver<SafekeeperTimelineInfo>,
+    // to unregister itself from shared state in Drop
+    registry: Registry,
+}
+
+impl Drop for Subscriber {
+    fn drop(&mut self) {
+        self.registry.unregister_subscriber(self);
+    }
+}
+
+// Private publisher state
+struct Publisher {
+    id: PubId,
+    registry: Registry,
+}
+
+impl Publisher {
+    // Send msg to relevant subscribers.
+    pub fn send_msg(&mut self, msg: &SafekeeperTimelineInfo) -> Result<(), Status> {
+        // send message to subscribers for everything
+        let shared_state = self.registry.shared_state.read();
+        // Err means there is no subscribers, it is fine.
+        shared_state.chan_to_all_subs.send(msg.clone()).ok();
+
+        // send message to per timeline subscribers
+        let ttid =
+            parse_proto_ttid(msg.tenant_timeline_id.as_ref().ok_or_else(|| {
+                Status::new(Code::InvalidArgument, "missing tenant_timeline_id")
+            })?)?;
+        if let Some(subs) = shared_state.chans_to_timeline_subs.get(&ttid) {
+            // Err can't happen here, as rx is destroyed after removing from the
+            // map the last subscriber.
+            subs.chan.send(msg.clone()).unwrap();
+        }
+        Ok(())
+    }
+}
+
+impl Drop for Publisher {
+    fn drop(&mut self) {
+        self.registry.unregister_publisher(self);
+    }
+}
+
+struct NeonBrokerImpl {
+    registry: Registry,
+}
+
+#[tonic::async_trait]
+impl NeonBroker for NeonBrokerImpl {
+    async fn publish_safekeeper_info(
+        &self,
+        request: Request<tonic::Streaming<SafekeeperTimelineInfo>>,
+    ) -> Result<Response<()>, Status> {
+        let mut publisher = self.registry.register_publisher();
+
+        let mut stream = request.into_inner();
+
+        loop {
+            match stream.next().await {
+                Some(Ok(msg)) => publisher.send_msg(&msg)?,
+                Some(Err(e)) => return Err(e), // grpc error from the stream
+                None => break,                 // closed stream
+            }
+        }
+
+        Ok(Response::new(()))
+    }
+
+    type SubscribeSafekeeperInfoStream =
+        Pin<Box<dyn Stream<Item = Result<SafekeeperTimelineInfo, Status>> + Send + 'static>>;
+
+    async fn subscribe_safekeeper_info(
+        &self,
+        request: Request<SubscribeSafekeeperInfoRequest>,
+    ) -> Result<Response<Self::SubscribeSafekeeperInfoStream>, Status> {
+        let proto_key = request
+            .into_inner()
+            .subscription_key
+            .ok_or_else(|| Status::new(Code::InvalidArgument, "missing subscription key"))?;
+        let sub_key = SubscriptionKey::from_proto_subscription_key(proto_key)?;
+        let mut subscriber = self.registry.register_subscriber(sub_key);
+
+        // transform rx into stream with item = Result, as method result demands
+        let output = async_stream::try_stream! {
+            loop {
+                match subscriber.sub_rx.recv().await {
+                    Ok(info) => yield info,
+                    Err(RecvError::Lagged(_skipped_msg)) => {
+                        // warn!("dropped {} messages, channel is full", skipped_msg);
+                    }
+                    Err(RecvError::Closed) => {
+                        // can't happen, we never drop the channel while there is a subscriber
+                        Err(Status::new(Code::Internal, "channel unexpectantly closed"))?;
+                    }
+                }
+            }
+        };
+
+        Ok(Response::new(
+            Box::pin(output) as Self::SubscribeSafekeeperInfoStream
+        ))
+    }
+}
+
+// We serve only metrics through http1.
+async fn http1_handler(
+    req: hyper::Request<hyper::body::Body>,
+) -> Result<hyper::Response<Body>, Infallible> {
+    let resp = match (req.method(), req.uri().path()) {
+        (&Method::GET, "/metrics") => {
+            let mut buffer = vec![];
+            let metrics = metrics::gather();
+            let encoder = TextEncoder::new();
+            encoder.encode(&metrics, &mut buffer).unwrap();
+
+            hyper::Response::builder()
+                .status(StatusCode::OK)
+                .header(CONTENT_TYPE, encoder.format_type())
+                .body(Body::from(buffer))
+                .unwrap()
+        }
+        _ => hyper::Response::builder()
+            .status(StatusCode::NOT_FOUND)
+            .body(Body::empty())
+            .unwrap(),
+    };
+    Ok(resp)
+}
+
+#[tokio::main]
+async fn main() -> Result<(), Box<dyn std::error::Error>> {
+    let args = Args::parse();
+
+    #[cfg(not(feature = "console"))] // tokio-console inits tracing_subscriber on its own
+    utils::logging::init(utils::logging::LogFormat::from_config(&args.log_format)?)?;
+    info!("version: {GIT_VERSION}");
+
+    #[cfg(feature = "console")]
+    console_subscriber::init();
+
+    let registry = Registry {
+        shared_state: Arc::new(RwLock::new(SharedState::new(args.chan_size))),
+        chan_size: args.chan_size,
+    };
+    let neon_broker_impl = NeonBrokerImpl {
+        registry: registry.clone(),
+    };
+    let neon_broker_server = NeonBrokerServer::new(neon_broker_impl);
+
+    info!("listening on {}", &args.listen_addr);
+
+    // grpc is served along with http1 for metrics on a single port, hence we
+    // don't use tonic's Server.
+    hyper::Server::bind(&args.listen_addr)
+        .http2_keep_alive_interval(Some(args.http2_keepalive_interval))
+        .serve(make_service_fn(move |_| {
+            let neon_broker_server_cloned = neon_broker_server.clone();
+            async move {
+                Ok::<_, Infallible>(service_fn(move |req| {
+                    // Technically this second clone is not needed, but consume
+                    // by async block is apparently unavoidable. BTW, error
+                    // message is enigmatic, see
+                    // https://github.com/rust-lang/rust/issues/68119
+                    //
+                    // We could get away without async block at all, but then we
+                    // need to resort to futures::Either to merge the result,
+                    // which doesn't caress an eye as well.
+                    let mut neon_broker_server_svc = neon_broker_server_cloned.clone();
+                    async move {
+                        if req.headers().get("content-type").map(|x| x.as_bytes())
+                            == Some(b"application/grpc")
+                        {
+                            let res_resp = neon_broker_server_svc.call(req).await;
+                            // Grpc and http1 handlers have slightly different
+                            // Response types: it is UnsyncBoxBody for the
+                            // former one (not sure why) and plain hyper::Body
+                            // for the latter. Both implement HttpBody though,
+                            // and EitherBody is used to merge them.
+                            res_resp.map(|resp| resp.map(EitherBody::Left))
+                        } else {
+                            let res_resp = http1_handler(req).await;
+                            res_resp.map(|resp| resp.map(EitherBody::Right))
+                        }
+                    }
+                }))
+            }
+        }))
+        .await?;
+    Ok(())
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use neon_broker::neon_broker_proto::TenantTimelineId as ProtoTenantTimelineId;
+    use tokio::sync::broadcast::error::TryRecvError;
+    use utils::id::{TenantId, TimelineId};
+
+    fn msg(timeline_id: Vec<u8>) -> SafekeeperTimelineInfo {
+        SafekeeperTimelineInfo {
+            safekeeper_id: 1,
+            tenant_timeline_id: Some(ProtoTenantTimelineId {
+                tenant_id: vec![0x00; 16],
+                timeline_id,
+            }),
+            last_log_term: 0,
+            flush_lsn: 1,
+            commit_lsn: 2,
+            backup_lsn: 3,
+            remote_consistent_lsn: 4,
+            peer_horizon_lsn: 5,
+            safekeeper_connstr: "neon-1-sk-1.local:7676".to_owned(),
+            local_start_lsn: 0,
+        }
+    }
+
+    fn tli_from_u64(i: u64) -> Vec<u8> {
+        let mut timeline_id = vec![0xFF; 8];
+        timeline_id.extend_from_slice(&i.to_be_bytes());
+        timeline_id
+    }
+
+    #[tokio::test]
+    async fn test_registry() {
+        let registry = Registry {
+            shared_state: Arc::new(RwLock::new(SharedState::new(16))),
+            chan_size: 16,
+        };
+
+        // subscribe to timeline 2
+        let ttid_2 = TenantTimelineId {
+            tenant_id: TenantId::from_slice(&[0x00; 16]).unwrap(),
+            timeline_id: TimelineId::from_slice(&tli_from_u64(2)).unwrap(),
+        };
+        let sub_key_2 = SubscriptionKey::Timeline(ttid_2);
+        let mut subscriber_2 = registry.register_subscriber(sub_key_2);
+        let mut subscriber_all = registry.register_subscriber(SubscriptionKey::All);
+
+        // send two messages with different keys
+        let msg_1 = msg(tli_from_u64(1));
+        let msg_2 = msg(tli_from_u64(2));
+        let mut publisher = registry.register_publisher();
+        publisher.send_msg(&msg_1).expect("failed to send msg");
+        publisher.send_msg(&msg_2).expect("failed to send msg");
+
+        // msg with key 2 should arrive to subscriber_2
+        assert!(
+            matches!(subscriber_2.sub_rx.try_recv(), Ok(msg) if matches!(msg.tenant_timeline_id.as_ref(), Some(ttid) if parse_proto_ttid(ttid).unwrap() == ttid_2))
+        );
+        // but nothing more
+        assert!(
+            matches!(subscriber_2.sub_rx.try_recv(), Err(err) if matches!(err, TryRecvError::Empty))
+        );
+
+        // subscriber_all should receive both messages
+        assert!(matches!(subscriber_all.sub_rx.try_recv(), Ok(..)));
+        assert!(matches!(subscriber_all.sub_rx.try_recv(), Ok(..)));
+        assert!(
+            matches!(subscriber_all.sub_rx.try_recv(), Err(err) if matches!(err, TryRecvError::Empty))
+        );
+    }
+}

broker/src/broker.rs

@@ -1,526 +0,0 @@
-//! Simple pub-sub based on grpc (tonic) and Tokio mpsc for storage nodes
-//! messaging. The main design goal is to avoid central synchronization during
-//! normal flow, resorting to it only when pub/sub change happens. Each
-//! subscriber holds mpsc for messages it sits on; tx end is sent to existing
-//! publishers and saved in shared state for new ones. Publishers maintain
-//! locally set of subscribers they stream messages to.
-//!
-//! Subscriptions to 1) single timeline 2) everything are possible. We could add
-//! subscription to set of timelines to save grpc streams, but testing shows
-//! many individual streams is also ok.
-//!
-//! Message is dropped if subscriber can't consume it, not affecting other
-//! subscribers.
-//!
-//! Only safekeeper message is supported, but it is not hard to add something
-//! else with templating.
-use std::collections::hash_map::Entry;
-use std::collections::HashMap;
-use std::fmt;
-use std::pin::Pin;
-use std::sync::{Arc, Mutex};
-use std::task::{Context, Poll};
-use std::time::Duration;
-
-use futures_core::Stream;
-use futures_util::StreamExt;
-use tokio::sync::mpsc::error::TrySendError;
-use tokio::sync::mpsc::{self, Receiver, Sender};
-use tokio::{select, time};
-use tokio_stream::wrappers::ReceiverStream;
-use tonic::Code;
-use tonic::{transport::Server, Request, Response, Status};
-
-use neon_broker_proto::neon_broker_server::{NeonBroker, NeonBrokerServer};
-use neon_broker_proto::subscribe_safekeeper_info_request::SubscriptionKey as ProtoSubscriptionKey;
-use neon_broker_proto::TenantTimelineId as ProtoTenantTimelineId;
-use neon_broker_proto::{Empty, SafekeeperTimelineInfo, SubscribeSafekeeperInfoRequest};
-use utils::id::{TenantId, TenantTimelineId, TimelineId};
-
-pub mod neon_broker_proto {
-    // The string specified here must match the proto package name.
-    // If you want to have a look at the generated code, it is at path similar to
-    // target/debug/build/neon_broker-0fde81d03bedc3b2/out/neon_broker.rs
-    tonic::include_proto!("neon_broker");
-}
-
-// Max size of the queue to the subscriber.
-const CHAN_SIZE: usize = 256;
-
-type PubId = u64; // id of publisher for registering in maps
-type SubId = u64; // id of subscriber for registering in maps
-
-#[derive(Copy, Clone)]
-enum SubscriptionKey {
-    All,
-    Timeline(TenantTimelineId),
-}
-
-impl SubscriptionKey {
-    // Parse protobuf subkey (protobuf doesn't have fixed size bytes, we get vectors).
-    pub fn from_proto_subscription_key(key: ProtoSubscriptionKey) -> Result<Self, Status> {
-        match key {
-            ProtoSubscriptionKey::All(_) => Ok(SubscriptionKey::All),
-            ProtoSubscriptionKey::TenantTimelineId(proto_ttid) => {
-                Ok(SubscriptionKey::Timeline(parse_proto_ttid(&proto_ttid)?))
-            }
-        }
-    }
-}
-
-// Subscriber id + tx end of the channel for messages to it.
-#[derive(Clone)]
-struct SubSender(SubId, Sender<SafekeeperTimelineInfo>);
-
-impl fmt::Debug for SubSender {
-    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
-        write!(f, "Subscription id {}", self.0)
-    }
-}
-
-// Announcements subscriber sends to publisher(s) asking it to stream to the
-// provided channel, or forget about it, releasing memory.
-#[derive(Clone)]
-enum SubAnnounce {
-    AddAll(Sender<SafekeeperTimelineInfo>), // add subscription to all timelines
-    AddTimeline(TenantTimelineId, SubSender), // add subsciption to the specific timeline
-    RemoveTimeline(TenantTimelineId, SubId), // remove subscription to the specific timeline
-                                            // RemoveAll is not needed as publisher will notice closed channel while
-                                            // trying to send the next message.
-}
-
-struct SharedState {
-    // Registered publishers. They sit on the rx end of these channels and
-    // receive through it tx handles of chans to subscribers.
-    //
-    // Note: publishers don't identify which keys they publish, so each
-    // publisher will receive channels to all subs and filter them before sending.
-    pub_txs: HashMap<PubId, Sender<SubAnnounce>>,
-    next_pub_id: PubId,
-    // Registered subscribers -- when publisher joins it walks over them,
-    // collecting txs to send messages.
-    subs_to_all: HashMap<SubId, Sender<SafekeeperTimelineInfo>>,
-    subs_to_timelines: HashMap<TenantTimelineId, Vec<SubSender>>,
-    next_sub_id: SubId,
-}
-
-// Utility func to remove subscription from the map
-fn remove_sub(
-    subs_to_timelines: &mut HashMap<TenantTimelineId, Vec<SubSender>>,
-    ttid: &TenantTimelineId,
-    sub_id: SubId,
-) {
-    if let Some(subsenders) = subs_to_timelines.get_mut(&ttid) {
-        subsenders.retain(|ss| ss.0 != sub_id);
-        if subsenders.len() == 0 {
-            subs_to_timelines.remove(&ttid);
-        }
-    }
-    // Note that subscription might be not here if subscriber task was aborted
-    // earlier than it managed to notify publisher about itself.
-}
-
-impl SharedState {
-    // Register new publisher.
-    pub fn register_publisher(&mut self, announce_tx: Sender<SubAnnounce>) -> PubId {
-        let pub_id = self.next_pub_id;
-        self.next_pub_id += 1;
-        assert!(!self.pub_txs.contains_key(&pub_id));
-        self.pub_txs.insert(pub_id, announce_tx);
-        pub_id
-    }
-
-    pub fn unregister_publisher(&mut self, pub_id: PubId) {
-        assert!(self.pub_txs.contains_key(&pub_id));
-        self.pub_txs.remove(&pub_id);
-    }
-
-    // Register new subscriber.
-    // Returns list of channels through which existing publishers must be notified
-    // about new subscriber; we can't do it here due to risk of deadlock.
-    pub fn register_subscriber(
-        &mut self,
-        sub_key: SubscriptionKey,
-        sub_tx: Sender<SafekeeperTimelineInfo>,
-    ) -> (SubId, Vec<Sender<SubAnnounce>>, SubAnnounce) {
-        let sub_id = self.next_sub_id;
-        self.next_sub_id += 1;
-        let announce = match sub_key {
-            SubscriptionKey::All => {
-                assert!(!self.subs_to_all.contains_key(&sub_id));
-                self.subs_to_all.insert(sub_id, sub_tx.clone());
-                SubAnnounce::AddAll(sub_tx)
-            }
-            SubscriptionKey::Timeline(ttid) => {
-                match self.subs_to_timelines.entry(ttid) {
-                    Entry::Occupied(mut o) => {
-                        let subsenders = o.get_mut();
-                        subsenders.push(SubSender(sub_id, sub_tx.clone()));
-                    }
-                    Entry::Vacant(v) => {
-                        v.insert(vec![SubSender(sub_id, sub_tx.clone())]);
-                    }
-                }
-                SubAnnounce::AddTimeline(ttid, SubSender(sub_id, sub_tx))
-            }
-        };
-        // Collect existing publishers to notify them after lock is released;
-        // TODO: the probability of channels being full here is tiny (publisher
-        // always blocks listening chan), we can try sending first and resort to
-        // cloning if needed.
-        //
-        // Deadlock is possible only if publisher tries to access shared state
-        // during its lifetime, i.e. we add maintenance of set of published
-        // tlis. Otherwise we can just await here (but lock must be replaced
-        // with Tokio one).
-        //
-        // We could also just error out if some chan is full, but that needs
-        // cleanup of incompleted job, and notifying publishers when unregistering
-        // is mandatory anyway.
-        (sub_id, self.pub_txs.values().cloned().collect(), announce)
-    }
-
-    // Unregister the subscriber. Similar to register_subscriber, returns list
-    // of channels through which publishers must be notified about the removal.
-    pub fn unregister_subscriber(
-        &mut self,
-        sub_id: SubId,
-        sub_key: SubscriptionKey,
-    ) -> Option<(Vec<Sender<SubAnnounce>>, SubAnnounce)> {
-        // We need to notify existing publishers only about per timeline
-        // subscriptions, 'all' kind is detected on its own through closed
-        // channels.
-        let announce = match sub_key {
-            SubscriptionKey::All => {
-                assert!(self.subs_to_all.contains_key(&sub_id));
-                self.subs_to_all.remove(&sub_id);
-                None
-            }
-            SubscriptionKey::Timeline(ref ttid) => {
-                remove_sub(&mut self.subs_to_timelines, ttid, sub_id);
-                Some(SubAnnounce::RemoveTimeline(*ttid, sub_id))
-            }
-        };
-        announce.map(|a| (self.pub_txs.values().cloned().collect(), a))
-    }
-
-    pub fn report(&mut self) {
-        println!(
-            "registered {} publishers, {} subs to all, {} subs to timelines",
-            self.pub_txs.len(),
-            self.subs_to_all.len(),
-            self.subs_to_timelines.len(),
-        );
-    }
-}
-
-// SharedState wrapper for post-locking operations (sending to pub_tx chans).
-#[derive(Clone)]
-struct Registry {
-    shared_state: Arc<Mutex<SharedState>>,
-}
-
-impl Registry {
-    // Register new publisher in shared state.
-    pub fn register_publisher(&self) -> Publisher {
-        let (announce_tx, announce_rx) = mpsc::channel(128);
-        let mut ss = self.shared_state.lock().unwrap();
-        let id = ss.register_publisher(announce_tx);
-        let (subs_to_all, subs_to_timelines) = (
-            ss.subs_to_all.values().cloned().collect(),
-            ss.subs_to_timelines.clone(),
-        );
-        drop(ss);
-        // println!("registered publisher {}", id);
-        Publisher {
-            id,
-            announce_rx: announce_rx.into(),
-            subs_to_all,
-            subs_to_timelines,
-            registry: self.clone(),
-        }
-    }
-
-    pub fn unregister_publisher(&self, publisher: &Publisher) {
-        self.shared_state
-            .lock()
-            .unwrap()
-            .unregister_publisher(publisher.id);
-        // println!("unregistered publisher {}", publisher.id);
-    }
-
-    // Register new subscriber in shared state.
-    pub async fn register_subscriber(&self, sub_key: SubscriptionKey) -> Subscriber {
-        let (tx, rx) = mpsc::channel(CHAN_SIZE);
-        let id;
-        let mut pub_txs;
-        let announce;
-        {
-            let mut ss = self.shared_state.lock().unwrap();
-            (id, pub_txs, announce) = ss.register_subscriber(sub_key, tx);
-        }
-        // Note: it is important to create Subscriber before .await. If client
-        // disconnects during await, which would terminate the Future we still
-        // need to run Subscriber's drop() which will unregister it from the
-        // shared state.
-        let subscriber = Subscriber {
-            id,
-            key: sub_key,
-            sub_rx: rx,
-            registry: self.clone(),
-        };
-        // Notify existing publishers about new subscriber.
-        for pub_tx in pub_txs.iter_mut() {
-            // Closed channel is fine; it means publisher has gone.
-            pub_tx.send(announce.clone()).await.ok();
-        }
-        // println!("registered subscriber {}", id);
-        subscriber
-    }
-
-    // Unregister the subscriber
-    pub fn unregister_subscriber(&self, sub: &Subscriber) {
-        let mut ss = self.shared_state.lock().unwrap();
-        let announce_pack = ss.unregister_subscriber(sub.id, sub.key);
-        drop(ss);
-        // Notify publishers about the removal. Apart from wanting to do it
-        // outside lock, here we also spin a task as Drop impl can't be async.
-        if let Some((mut pub_txs, announce)) = announce_pack {
-            tokio::spawn(async move {
-                for pub_tx in pub_txs.iter_mut() {
-                    // Closed channel is fine; it means publisher has gone.
-                    pub_tx.send(announce.clone()).await.ok();
-                }
-            });
-        }
-        // println!("unregistered subscriber {}", sub.id);
-    }
-
-    pub async fn report(&self) {
-        let mut interval = time::interval(Duration::from_millis(1000));
-        loop {
-            interval.tick().await;
-            self.shared_state.lock().unwrap().report();
-        }
-    }
-}
-
-// Private subscriber state.
-struct Subscriber {
-    id: SubId,
-    key: SubscriptionKey,
-    // Subscriber receives messages from publishers here.
-    sub_rx: Receiver<SafekeeperTimelineInfo>,
-    // to unregister itself from shared state in Drop
-    registry: Registry,
-}
-
-impl Drop for Subscriber {
-    fn drop(&mut self) {
-        self.registry.unregister_subscriber(self);
-    }
-}
-
-// Private publisher state
-struct Publisher {
-    id: PubId,
-    // new subscribers request to send (or stop sending) msgs them here.
-    // It could be just Receiver, but weirdly it doesn't implement futures_core Stream directly.
-    announce_rx: ReceiverStream<SubAnnounce>,
-    subs_to_all: Vec<Sender<SafekeeperTimelineInfo>>,
-    subs_to_timelines: HashMap<TenantTimelineId, Vec<SubSender>>,
-    // to unregister itself from shared state in Drop
-    registry: Registry,
-}
-
-impl Publisher {
-    // Send msg to relevant subscribers.
-    pub fn send_msg(&mut self, msg: &SafekeeperTimelineInfo) -> Result<(), Status> {
-        // send message to subscribers for everything
-        let mut cleanup_subs_to_all = false;
-        for sub in self.subs_to_all.iter() {
-            match sub.try_send(msg.clone()) {
-                Err(TrySendError::Full(_)) => {
-                    // println!("dropping message, channel is full");
-                }
-                Err(TrySendError::Closed(_)) => {
-                    cleanup_subs_to_all = true;
-                }
-                _ => (),
-            }
-        }
-        // some channels got closed (subscriber gone), remove them
-        if cleanup_subs_to_all {
-            self.subs_to_all.retain(|tx| !tx.is_closed());
-        }
-
-        // send message to per timeline subscribers
-        let ttid = parse_proto_ttid(msg.tenant_timeline_id.as_ref().ok_or(Status::new(
-            Code::InvalidArgument,
-            "missing tenant_timeline_id",
-        ))?)?;
-        if let Some(subs) = self.subs_to_timelines.get(&ttid) {
-            for tx in subs.iter().map(|sub_sender| &sub_sender.1) {
-                if let Err(TrySendError::Full(_)) = tx.try_send(msg.clone()) {
-                    // println!("dropping message, channel is full");
-                }
-                // closed channel is ignored here; we will be notified and remove it soon
-            }
-        }
-        Ok(())
-    }
-
-    // Add/remove subscriber according to sub_announce.
-    pub fn update_sub(&mut self, sub_announce: SubAnnounce) {
-        match sub_announce {
-            SubAnnounce::AddAll(tx) => self.subs_to_all.push(tx),
-            SubAnnounce::AddTimeline(ttid, sub_sender) => {
-                match self.subs_to_timelines.entry(ttid) {
-                    Entry::Occupied(mut o) => {
-                        let subsenders = o.get_mut();
-                        subsenders.push(sub_sender);
-                    }
-                    Entry::Vacant(v) => {
-                        v.insert(vec![sub_sender]);
-                    }
-                }
-            }
-            SubAnnounce::RemoveTimeline(ref ttid, sub_id) => {
-                remove_sub(&mut self.subs_to_timelines, ttid, sub_id);
-            }
-        }
-    }
-}
-
-impl Drop for Publisher {
-    fn drop(&mut self) {
-        self.registry.unregister_publisher(self);
-    }
-}
-
-struct NeonBrokerService {
-    registry: Registry,
-}
-
-#[tonic::async_trait]
-impl NeonBroker for NeonBrokerService {
-    async fn publish_safekeeper_info(
-        &self,
-        request: Request<tonic::Streaming<SafekeeperTimelineInfo>>,
-    ) -> Result<Response<Empty>, Status> {
-        let mut publisher = self.registry.register_publisher();
-
-        let mut stream = request.into_inner();
-
-        loop {
-            select! {
-                msg = stream.next() => {
-                    match msg {
-                        Some(Ok(msg)) => {publisher.send_msg(&msg)?;},
-                        Some(Err(e)) => {return Err(e);}, // grpc error from the stream
-                        None => {break;} // closed stream
-                    }
-                }
-                Some(announce) = publisher.announce_rx.next() => {
-                    publisher.update_sub(announce);
-                }
-            }
-        }
-
-        Ok(Response::new(Empty {}))
-    }
-
-    type SubscribeSafekeeperInfoStream =
-        Pin<Box<dyn Stream<Item = Result<SafekeeperTimelineInfo, Status>> + Send + 'static>>;
-
-    async fn subscribe_safekeeper_info(
-        &self,
-        request: Request<SubscribeSafekeeperInfoRequest>,
-    ) -> Result<Response<Self::SubscribeSafekeeperInfoStream>, Status> {
-        let proto_key = request.into_inner().subscription_key.ok_or(Status::new(
-            Code::InvalidArgument,
-            "missing subscription key",
-        ))?;
-        let sub_key = SubscriptionKey::from_proto_subscription_key(proto_key)?;
-        let mut subscriber = self.registry.register_subscriber(sub_key).await;
-
-        // transform rx into stream with item = Result, as method result demands
-        let output = async_stream::try_stream! {
-            while let Some(info) = subscriber.sub_rx.recv().await {
-                    yield info
-                }
-
-            // internal generator
-            // let _ = subscriber.sub_rx.try_recv().ok();
-            // let mut counter = 0;
-            // loop {
-            // let info = SafekeeperTimelineInfo {
-            //     safekeeper_id: 1,
-            //     tenant_timeline_id: Some(ProtoTenantTimelineId {
-            //         tenant_id: vec![0xFF; 16],
-            //         timeline_id: vec![0xFF; 16],
-            //         // timeline_id: tli_from_u64(counter),
-            //     }),
-            //     last_log_term: 0,
-            //     flush_lsn: counter,
-            //     commit_lsn: 2,
-            //     backup_lsn: 3,
-            //     remote_consistent_lsn: 4,
-            //     peer_horizon_lsn: 5,
-            //     safekeeper_connstr: "zenith-1-sk-1.local:7676".to_owned(),
-            // };
-            // counter += 1;
-            // yield info;
-            // }
-        };
-
-        Ok(Response::new(
-            Box::pin(output) as Self::SubscribeSafekeeperInfoStream
-        ))
-    }
-}
-
-#[tokio::main]
-async fn main() -> Result<(), Box<dyn std::error::Error>> {
-    // console_subscriber::init();
-
-    let addr = "[::1]:50051".parse()?;
-    let registry = Registry {
-        shared_state: Arc::new(Mutex::new(SharedState {
-            pub_txs: HashMap::new(),
-            next_pub_id: 0,
-            subs_to_all: HashMap::new(),
-            subs_to_timelines: HashMap::new(),
-            next_sub_id: 0,
-        })),
-    };
-    let neon_broker_service = NeonBrokerService {
-        registry: registry.clone(),
-    };
-
-    tokio::spawn(async move { registry.report().await });
-
-    Server::builder()
-        .http2_keepalive_interval(Some(Duration::from_millis(5000)))
-        .add_service(NeonBrokerServer::new(neon_broker_service))
-        .serve(addr)
-        .await?;
-
-    Ok(())
-}
-
-// parse variable length bytes from protobuf
-fn parse_proto_ttid(proto_ttid: &ProtoTenantTimelineId) -> Result<TenantTimelineId, Status> {
-    let tenant_id = TenantId::from_vec(&proto_ttid.tenant_id)
-        .map_err(|e| Status::new(Code::InvalidArgument, format!("malformed tenant_id: {}", e)))?;
-    let timeline_id = TimelineId::from_vec(&proto_ttid.timeline_id).map_err(|e| {
-        Status::new(
-            Code::InvalidArgument,
-            format!("malformed timeline_id: {}", e),
-        )
-    })?;
-    Ok(TenantTimelineId {
-        tenant_id,
-        timeline_id,
-    })
-}

broker/src/lib.rs

@@ -0,0 +1,108 @@
+use hyper::body::HttpBody;
+use std::pin::Pin;
+use std::task::{Context, Poll};
+use tonic::codegen::StdError;
+use tonic::{transport::Channel, Code, Status};
+use utils::id::{TenantId, TenantTimelineId, TimelineId};
+
+use neon_broker_proto::{
+    neon_broker_client::NeonBrokerClient, TenantTimelineId as ProtoTenantTimelineId,
+};
+
+// Code generated by protobuf.
+pub mod neon_broker_proto {
+    include!("../proto/neon_broker.rs");
+}
+
+pub mod metrics;
+
+// Re-exports to avoid direct tonic dependency in user crates.
+pub use tonic::Request;
+pub use tonic::Streaming;
+
+pub const DEFAULT_LISTEN_ADDR: &str = "127.0.0.1:50051";
+
+// NeonBrokerClient charged with tonic provided Channel transport; helps to
+// avoid depending on tonic directly in user crates.
+pub type NeonBrokerClientChannel = NeonBrokerClient<Channel>;
+
+impl NeonBrokerClient<Channel> {
+    /// Create a new client to the given endpoint, but don't actually connect until the first request.
+    pub async fn connect_lazy<D>(dst: D) -> Result<Self, tonic::transport::Error>
+    where
+        D: std::convert::TryInto<tonic::transport::Endpoint>,
+        D::Error: Into<StdError>,
+    {
+        let conn = tonic::transport::Endpoint::new(dst)?.connect_lazy();
+        Ok(Self::new(conn))
+    }
+}
+
+// parse variable length bytes from protobuf
+pub fn parse_proto_ttid(proto_ttid: &ProtoTenantTimelineId) -> Result<TenantTimelineId, Status> {
+    let tenant_id = TenantId::from_slice(&proto_ttid.tenant_id)
+        .map_err(|e| Status::new(Code::InvalidArgument, format!("malformed tenant_id: {}", e)))?;
+    let timeline_id = TimelineId::from_slice(&proto_ttid.timeline_id).map_err(|e| {
+        Status::new(
+            Code::InvalidArgument,
+            format!("malformed timeline_id: {}", e),
+        )
+    })?;
+    Ok(TenantTimelineId {
+        tenant_id,
+        timeline_id,
+    })
+}
+
+// These several usages don't justify anyhow dependency, though it would work as
+// well.
+type AnyError = Box<dyn std::error::Error + Send + Sync + 'static>;
+
+// Provides impl HttpBody for two different types implementing it. Inspired by
+// https://github.com/hyperium/tonic/blob/master/examples/src/hyper_warp/server.rs
+pub enum EitherBody<A, B> {
+    Left(A),
+    Right(B),
+}
+
+impl<A, B> HttpBody for EitherBody<A, B>
+where
+    A: HttpBody + Send + Unpin,
+    B: HttpBody<Data = A::Data> + Send + Unpin,
+    A::Error: Into<AnyError>,
+    B::Error: Into<AnyError>,
+{
+    type Data = A::Data;
+    type Error = Box<dyn std::error::Error + Send + Sync + 'static>;
+
+    fn is_end_stream(&self) -> bool {
+        match self {
+            EitherBody::Left(b) => b.is_end_stream(),
+            EitherBody::Right(b) => b.is_end_stream(),
+        }
+    }
+
+    fn poll_data(
+        self: Pin<&mut Self>,
+        cx: &mut Context<'_>,
+    ) -> Poll<Option<Result<Self::Data, Self::Error>>> {
+        match self.get_mut() {
+            EitherBody::Left(b) => Pin::new(b).poll_data(cx).map(map_option_err),
+            EitherBody::Right(b) => Pin::new(b).poll_data(cx).map(map_option_err),
+        }
+    }
+
+    fn poll_trailers(
+        self: Pin<&mut Self>,
+        cx: &mut Context<'_>,
+    ) -> Poll<Result<Option<hyper::HeaderMap>, Self::Error>> {
+        match self.get_mut() {
+            EitherBody::Left(b) => Pin::new(b).poll_trailers(cx).map_err(Into::into),
+            EitherBody::Right(b) => Pin::new(b).poll_trailers(cx).map_err(Into::into),
+        }
+    }
+}
+
+fn map_option_err<T, U: Into<AnyError>>(err: Option<Result<T, U>>) -> Option<Result<T, AnyError>> {
+    err.map(|e| e.map_err(Into::into))
+}

broker/src/metrics.rs

@@ -0,0 +1,25 @@
+//! Broker metrics.
+
+use metrics::{register_int_gauge, IntGauge};
+use once_cell::sync::Lazy;
+
+pub static NUM_PUBS: Lazy<IntGauge> = Lazy::new(|| {
+    register_int_gauge!("broker_active_publishers", "Number of publications")
+        .expect("Failed to register metric")
+});
+
+pub static NUM_SUBS_TIMELINE: Lazy<IntGauge> = Lazy::new(|| {
+    register_int_gauge!(
+        "broker_per_timeline_active_subscribers",
+        "Number of subsciptions to particular tenant timeline id"
+    )
+    .expect("Failed to register metric")
+});
+
+pub static NUM_SUBS_ALL: Lazy<IntGauge> = Lazy::new(|| {
+    register_int_gauge!(
+        "broker_all_keys_active_subscribers",
+        "Number of subsciptions to all keys"
+    )
+    .expect("Failed to register metric")
+});

compute_tools/Cargo.toml

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

compute_tools/src/bin/compute_ctl.rs

@@ -51,53 +51,19 @@ fn main() -> Result<()> {
     // TODO: re-use `utils::logging` later
     init_logger(DEFAULT_LOG_LEVEL)?;
 
-    // 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 matches = cli().get_matches();
 
-    let pgdata = matches.value_of("pgdata").expect("PGDATA path is required");
+    let pgdata = matches
+        .get_one::<String>("pgdata")
+        .expect("PGDATA path is required");
     let connstr = matches
-        .value_of("connstr")
+        .get_one::<String>("connstr")
         .expect("Postgres connection string is required");
-    let spec = matches.value_of("spec");
-    let spec_path = matches.value_of("spec-path");
+    let spec = matches.get_one::<String>("spec");
+    let spec_path = matches.get_one::<String>("spec-path");
 
     // Try to use just 'postgres' if no path is provided
-    let pgbin = matches.value_of("pgbin").unwrap_or("postgres");
+    let pgbin = matches.get_one::<String>("pgbin").unwrap();
 
     let spec: ComputeSpec = match spec {
         // First, try to get cluster spec from the cli argument
@@ -173,3 +139,48 @@ 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

@@ -8,11 +8,10 @@ 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
@@ -66,7 +65,7 @@ impl GenericOption {
             let name = match self.name.as_str() {
                 "safekeepers" => "neon.safekeepers",
                 "wal_acceptor_reconnect" => "neon.safekeeper_reconnect_timeout",
-                "wal_acceptor_connect_timeout" => "neon.safekeeper_connect_timeout",
+                "wal_acceptor_connection_timeout" => "neon.safekeeper_connection_timeout",
                 it => it,
             };
 
@@ -169,7 +168,7 @@ 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.quote())
+        write!(params, " OWNER {}", &self.owner.pg_quote())
             .expect("String is documented to not to error during write operations");
 
         params
@@ -180,18 +179,17 @@ impl Database {
 /// intended to be used for DB / role names.
 pub type PgIdent = 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;
+/// 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;
 }
 
-impl PgQuote for PgIdent {
+impl Escaping 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 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 pg_quote(&self) -> String {
         let result = format!("\"{}\"", self.replace('"', "\"\""));
         result
     }

compute_tools/src/spec.rs

@@ -1,7 +1,9 @@
 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;
 
@@ -115,8 +117,8 @@ 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.quote(),
-                            new_name.quote()
+                            op.name.pg_quote(),
+                            new_name.pg_quote()
                         );
 
                         warn!("renaming role '{}' to '{}'", op.name, new_name);
@@ -162,7 +164,7 @@ pub fn handle_roles(spec: &ComputeSpec, client: &mut Client) -> Result<()> {
             }
 
             if update_role {
-                let mut query: String = format!("ALTER ROLE {} ", name.quote());
+                let mut query: String = format!("ALTER ROLE {} ", name.pg_quote());
                 info_print!(" -> update");
 
                 query.push_str(&role.to_pg_options());
@@ -170,7 +172,7 @@ pub fn handle_roles(spec: &ComputeSpec, client: &mut Client) -> Result<()> {
             }
         } else {
             info!("role name: '{}'", &name);
-            let mut query: String = format!("CREATE ROLE {} ", name.quote());
+            let mut query: String = format!("CREATE ROLE {} ", name.pg_quote());
             info!("role create query: '{}'", &query);
             info_print!(" -> create");
 
@@ -179,7 +181,7 @@ 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.quote()
+                name.pg_quote()
             );
             xact.execute(grant_query.as_str(), &[])?;
             info!("role grant query: '{}'", &grant_query);
@@ -215,7 +217,7 @@ 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.quote());
+                let query: String = format!("DROP ROLE IF EXISTS {}", &op.name.pg_quote());
 
                 warn!("deleting role '{}'", &op.name);
                 xact.execute(query.as_str(), &[])?;
@@ -230,17 +232,16 @@ 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 connstr = node.connstr.clone();
-            // database name is always the last and the only component of the path
-            connstr.set_path(&db.name);
+            let mut conf = Config::from_str(node.connstr.as_str())?;
+            conf.dbname(&db.name);
 
-            let mut client = Client::connect(connstr.as_str(), NoTls)?;
+            let mut client = conf.connect(NoTls)?;
 
             // This will reassign all dependent objects to the db owner
             let reassign_query = format!(
                 "REASSIGN OWNED BY {} TO {}",
-                role_name.quote(),
-                db.owner.quote()
+                role_name.pg_quote(),
+                db.owner.pg_quote()
             );
             info!(
                 "reassigning objects owned by '{}' in db '{}' to '{}'",
@@ -249,7 +250,7 @@ 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.quote());
+            let drop_query = format!("DROP OWNED BY {}", role_name.pg_quote());
             client.simple_query(&drop_query)?;
         }
     }
@@ -279,7 +280,7 @@ 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.quote());
+                    let query: String = format!("DROP DATABASE IF EXISTS {}", &op.name.pg_quote());
 
                     warn!("deleting database '{}'", &op.name);
                     client.execute(query.as_str(), &[])?;
@@ -291,8 +292,8 @@ 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.quote(),
-                            new_name.quote()
+                            op.name.pg_quote(),
+                            new_name.pg_quote()
                         );
 
                         warn!("renaming database '{}' to '{}'", op.name, new_name);
@@ -320,7 +321,7 @@ 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.quote()
+                db.owner.pg_quote()
             } else {
                 db.owner.clone()
             };
@@ -328,15 +329,15 @@ pub fn handle_databases(spec: &ComputeSpec, client: &mut Client) -> Result<()> {
             if new_owner != r.owner {
                 let query: String = format!(
                     "ALTER DATABASE {} OWNER TO {}",
-                    name.quote(),
-                    db.owner.quote()
+                    name.pg_quote(),
+                    db.owner.pg_quote()
                 );
                 info_print!(" -> update");
 
                 client.execute(query.as_str(), &[])?;
             }
         } else {
-            let mut query: String = format!("CREATE DATABASE {} ", name.quote());
+            let mut query: String = format!("CREATE DATABASE {} ", name.pg_quote());
             info_print!(" -> create");
 
             query.push_str(&db.to_pg_options());
@@ -366,7 +367,7 @@ pub fn handle_grants(node: &ComputeNode, client: &mut Client) -> Result<()> {
         .cluster
         .roles
         .iter()
-        .map(|r| r.name.quote())
+        .map(|r| r.name.pg_quote())
         .collect::<Vec<_>>();
 
     for db in &spec.cluster.databases {
@@ -374,7 +375,7 @@ pub fn handle_grants(node: &ComputeNode, client: &mut Client) -> Result<()> {
 
         let query: String = format!(
             "GRANT CREATE ON DATABASE {} TO {}",
-            dbname.quote(),
+            dbname.pg_quote(),
             roles.join(", ")
         );
         info!("grant query {}", &query);
@@ -385,12 +386,11 @@ 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 {
-        // database name is always the last and the only component of the path
-        db_connstr.set_path(&db.name);
+        let mut conf = Config::from_str(node.connstr.as_str())?;
+        conf.dbname(&db.name);
 
-        let mut db_client = Client::connect(db_connstr.as_str(), NoTls)?;
+        let mut db_client = conf.connect(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`
@@ -419,9 +419,36 @@ pub fn handle_grants(node: &ComputeNode, client: &mut Client) -> Result<()> {
                     END IF;\n\
                 END\n\
             $$;",
-            db.owner.quote()
+            db.owner.pg_quote()
         );
         db_client.simple_query(&alter_query)?;
+
+        // Explicitly grant CREATE ON SCHEMA PUBLIC to the web_access user.
+        // This is needed because since postgres 15 this privilege is removed by default.
+        let grant_query = "DO $$\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 quote_ident() {
+    fn ident_pg_quote() {
         let ident: PgIdent = PgIdent::from("\"name\";\\n select 1;");
 
-        assert_eq!(ident.quote(), "\"\"\"name\"\";\\n select 1;\"");
+        assert_eq!(ident.pg_quote(), "\"\"\"name\"\";\\n select 1;\"");
     }
 }

control_plane/Cargo.toml

@@ -4,20 +4,21 @@ version = "0.1.0"
 edition = "2021"
 
 [dependencies]
-clap = "3.0"
+anyhow = "1.0"
+clap = "4.0"
 comfy-table = "6.1"
 git-version = "0.3.5"
-tar = "0.4.38"
+nix = "0.25"
+once_cell = "1.13.0"
 postgres = { git = "https://github.com/neondatabase/rust-postgres.git", rev = "d052ee8b86fff9897c77b0fe89ea9daba0e1fa38" }
+regex = "1"
+reqwest = { version = "0.11", default-features = false, features = ["blocking", "json", "rustls-tls"] }
 serde = { version = "1.0", features = ["derive"] }
 serde_with = "2.0"
-toml = "0.5"
-once_cell = "1.13.0"
-regex = "1"
-anyhow = "1.0"
+tar = "0.4.38"
 thiserror = "1"
-nix = "0.25"
-reqwest = { version = "0.11", default-features = false, features = ["blocking", "json", "rustls-tls"] }
+toml = "0.5"
+url = "2.2.2"
 
 # Note: Do not directly depend on pageserver or safekeeper; use pageserver_api or safekeeper_api
 # instead, so that recompile times are better.

control_plane/src/background_process.rs

@@ -0,0 +1,264 @@
+//! Spawns and kills background processes that are needed by Neon CLI.
+//! Applies common set-up such as log and pid files (if needed) to every process.
+//!
+//! Neon CLI does not run in background, so it needs to store the information about
+//! spawned processes, which it does in this module.
+//! We do that by storing the pid of the process in the "${process_name}.pid" file.
+//! The pid file can be created by the process itself
+//! (Neon storage binaries do that and also ensure that a lock is taken onto that file)
+//! or we create such file after starting the process
+//! (non-Neon binaries don't necessarily follow our pidfile conventions).
+//! The pid stored in the file is later used to stop the service.
+//!
+//! See [`lock_file`] module for more info.
+
+use std::ffi::OsStr;
+use std::io::Write;
+use std::path::Path;
+use std::process::{Child, Command};
+use std::time::Duration;
+use std::{fs, io, thread};
+
+use anyhow::{anyhow, bail, Context, Result};
+use nix::errno::Errno;
+use nix::sys::signal::{kill, Signal};
+use nix::unistd::Pid;
+
+use utils::lock_file;
+
+const RETRIES: u32 = 15;
+const RETRY_TIMEOUT_MILLIS: u64 = 500;
+
+/// Argument to `start_process`, to indicate whether it should create pidfile or if the process creates
+/// it itself.
+pub enum InitialPidFile<'t> {
+    /// Create a pidfile, to allow future CLI invocations to manipulate the process.
+    Create(&'t Path),
+    /// The process will create the pidfile itself, need to wait for that event.
+    Expect(&'t Path),
+}
+
+/// Start a background child process using the parameters given.
+pub fn start_process<F, S: AsRef<OsStr>>(
+    process_name: &str,
+    datadir: &Path,
+    command: &Path,
+    args: &[S],
+    initial_pid_file: InitialPidFile,
+    process_status_check: F,
+) -> anyhow::Result<Child>
+where
+    F: Fn() -> anyhow::Result<bool>,
+{
+    let log_path = datadir.join(format!("{process_name}.log"));
+    let process_log_file = fs::OpenOptions::new()
+        .create(true)
+        .write(true)
+        .append(true)
+        .open(&log_path)
+        .with_context(|| {
+            format!("Could not open {process_name} log file {log_path:?} for writing")
+        })?;
+    let same_file_for_stderr = process_log_file.try_clone().with_context(|| {
+        format!("Could not reuse {process_name} log file {log_path:?} for writing stderr")
+    })?;
+
+    let mut command = Command::new(command);
+    let background_command = command
+        .stdout(process_log_file)
+        .stderr(same_file_for_stderr)
+        .args(args);
+    let filled_cmd = fill_aws_secrets_vars(fill_rust_env_vars(background_command));
+
+    let mut spawned_process = filled_cmd.spawn().with_context(|| {
+        format!("Could not spawn {process_name}, see console output and log files for details.")
+    })?;
+    let pid = spawned_process.id();
+    let pid = Pid::from_raw(
+        i32::try_from(pid)
+            .with_context(|| format!("Subprocess {process_name} has invalid pid {pid}"))?,
+    );
+
+    let pid_file_to_check = match initial_pid_file {
+        InitialPidFile::Create(target_pid_file_path) => {
+            match lock_file::create_lock_file(target_pid_file_path, pid.to_string()) {
+                lock_file::LockCreationResult::Created { .. } => {
+                    // We use "lock" file here only to create the pid file. The lock on the pidfile will be dropped as soon
+                    // as this CLI invocation exits, so it's a bit useless, but doesn't any harm either.
+                }
+                lock_file::LockCreationResult::AlreadyLocked { .. } => {
+                    anyhow::bail!("Cannot write pid file for {process_name} at path {target_pid_file_path:?}: file is already locked by another process")
+                }
+                lock_file::LockCreationResult::CreationFailed(e) => {
+                    return Err(e.context(format!(
+                    "Failed to create pid file for {process_name} at path {target_pid_file_path:?}"
+                )))
+                }
+            }
+            None
+        }
+        InitialPidFile::Expect(pid_file_path) => Some(pid_file_path),
+    };
+
+    for retries in 0..RETRIES {
+        match process_started(pid, pid_file_to_check, &process_status_check) {
+            Ok(true) => {
+                println!("\n{process_name} started, pid: {pid}");
+                return Ok(spawned_process);
+            }
+            Ok(false) => {
+                if retries < 5 {
+                    print!(".");
+                    io::stdout().flush().unwrap();
+                } else {
+                    if retries == 5 {
+                        println!() // put a line break after dots for second message
+                    }
+                    println!("{process_name} has not started yet, retrying ({retries})...");
+                }
+                thread::sleep(Duration::from_millis(RETRY_TIMEOUT_MILLIS));
+            }
+            Err(e) => {
+                println!("{process_name} failed to start: {e:#}");
+                if let Err(e) = spawned_process.kill() {
+                    println!("Could not stop {process_name} subprocess: {e:#}")
+                };
+                return Err(e);
+            }
+        }
+    }
+    anyhow::bail!("{process_name} could not start in {RETRIES} attempts");
+}
+
+/// Stops the process, using the pid file given. Returns Ok also if the process is already not running.
+pub fn stop_process(immediate: bool, process_name: &str, pid_file: &Path) -> anyhow::Result<()> {
+    if !pid_file.exists() {
+        println!("{process_name} is already stopped: no pid file {pid_file:?} is present");
+        return Ok(());
+    }
+    let pid = read_pidfile(pid_file)?;
+
+    let sig = if immediate {
+        print!("Stopping {process_name} with pid {pid} immediately..");
+        Signal::SIGQUIT
+    } else {
+        print!("Stopping {process_name} with pid {pid} gracefully..");
+        Signal::SIGTERM
+    };
+    io::stdout().flush().unwrap();
+    match kill(pid, sig) {
+        Ok(()) => (),
+        Err(Errno::ESRCH) => {
+            println!(
+                "{process_name} with pid {pid} does not exist, but a pid file {pid_file:?} was found"
+            );
+            return Ok(());
+        }
+        Err(e) => anyhow::bail!("Failed to send signal to {process_name} with pid {pid}: {e}"),
+    }
+
+    // Wait until process is gone
+    for _ in 0..RETRIES {
+        match process_has_stopped(pid) {
+            Ok(true) => {
+                println!("\n{process_name} stopped");
+                if let Err(e) = fs::remove_file(pid_file) {
+                    if e.kind() != io::ErrorKind::NotFound {
+                        eprintln!("Failed to remove pid file {pid_file:?} after stopping the process: {e:#}");
+                    }
+                }
+                return Ok(());
+            }
+            Ok(false) => {
+                print!(".");
+                io::stdout().flush().unwrap();
+                thread::sleep(Duration::from_secs(1))
+            }
+            Err(e) => {
+                println!("{process_name} with pid {pid} failed to stop: {e:#}");
+                return Err(e);
+            }
+        }
+    }
+
+    anyhow::bail!("{process_name} with pid {pid} failed to stop in {RETRIES} attempts");
+}
+
+fn fill_rust_env_vars(cmd: &mut Command) -> &mut Command {
+    let mut filled_cmd = cmd.env_clear().env("RUST_BACKTRACE", "1");
+
+    let var = "LLVM_PROFILE_FILE";
+    if let Some(val) = std::env::var_os(var) {
+        filled_cmd = filled_cmd.env(var, val);
+    }
+
+    const RUST_LOG_KEY: &str = "RUST_LOG";
+    if let Ok(rust_log_value) = std::env::var(RUST_LOG_KEY) {
+        filled_cmd.env(RUST_LOG_KEY, rust_log_value)
+    } else {
+        filled_cmd
+    }
+}
+
+fn fill_aws_secrets_vars(mut cmd: &mut Command) -> &mut Command {
+    for env_key in [
+        "AWS_ACCESS_KEY_ID",
+        "AWS_SECRET_ACCESS_KEY",
+        "AWS_SESSION_TOKEN",
+    ] {
+        if let Ok(value) = std::env::var(env_key) {
+            cmd = cmd.env(env_key, value);
+        }
+    }
+    cmd
+}
+
+fn process_started<F>(
+    pid: Pid,
+    pid_file_to_check: Option<&Path>,
+    status_check: &F,
+) -> anyhow::Result<bool>
+where
+    F: Fn() -> anyhow::Result<bool>,
+{
+    match status_check() {
+        Ok(true) => match pid_file_to_check {
+            Some(pid_file_path) => {
+                if pid_file_path.exists() {
+                    let pid_in_file = read_pidfile(pid_file_path)?;
+                    Ok(pid_in_file == pid)
+                } else {
+                    Ok(false)
+                }
+            }
+            None => Ok(true),
+        },
+        Ok(false) => Ok(false),
+        Err(e) => anyhow::bail!("process failed to start: {e}"),
+    }
+}
+
+/// Read a PID file
+///
+/// We expect a file that contains a single integer.
+fn read_pidfile(pidfile: &Path) -> Result<Pid> {
+    let pid_str = fs::read_to_string(pidfile)
+        .with_context(|| format!("failed to read pidfile {pidfile:?}"))?;
+    let pid: i32 = pid_str
+        .parse()
+        .map_err(|_| anyhow!("failed to parse pidfile {pidfile:?}"))?;
+    if pid < 1 {
+        bail!("pidfile {pidfile:?} contained bad value '{pid}'");
+    }
+    Ok(Pid::from_raw(pid))
+}
+
+fn process_has_stopped(pid: Pid) -> anyhow::Result<bool> {
+    match kill(pid, None) {
+        // Process exists, keep waiting
+        Ok(_) => Ok(false),
+        // Process not found, we're done
+        Err(Errno::ESRCH) => Ok(true),
+        Err(err) => anyhow::bail!("Failed to send signal to process with pid {pid}: {err}"),
+    }
+}

control_plane/src/bin/neon_local.rs

@@ -6,11 +6,11 @@
 //! rely on `neon_local` to set up the environment for each test.
 //!
 use anyhow::{anyhow, bail, Context, Result};
-use clap::{App, AppSettings, Arg, ArgMatches};
+use clap::{value_parser, Arg, ArgAction, ArgMatches, Command};
 use control_plane::compute::ComputeControlPlane;
 use control_plane::local_env::{EtcdBroker, LocalEnv};
+use control_plane::pageserver::PageServerNode;
 use control_plane::safekeeper::SafekeeperNode;
-use control_plane::storage::PageServerNode;
 use control_plane::{etcd, local_env};
 use pageserver_api::models::TimelineInfo;
 use pageserver_api::{
@@ -85,212 +85,7 @@ struct TimelineTreeEl {
 //   * Providing CLI api to the pageserver
 //   * TODO: export/import to/from usual postgres
 fn main() -> Result<()> {
-    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 matches = cli().get_matches();
 
     let (sub_name, sub_args) = match matches.subcommand() {
         Some(subcommand_data) => subcommand_data,
@@ -475,16 +270,16 @@ 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
-        .value_of("tenant-id")
-        .map(TenantId::from_str)
+        .get_one::<String>("tenant-id")
+        .map(|tenant_id| TenantId::from_str(tenant_id))
         .transpose()
         .context("Failed to parse tenant id from the argument string")
 }
 
 fn parse_timeline_id(sub_match: &ArgMatches) -> anyhow::Result<Option<TimelineId>> {
     sub_match
-        .value_of("timeline-id")
-        .map(TimelineId::from_str)
+        .get_one::<String>("timeline-id")
+        .map(|timeline_id| TimelineId::from_str(timeline_id))
         .transpose()
         .context("Failed to parse timeline id from the argument string")
 }
@@ -493,19 +288,22 @@ 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.value_of("config") {
+    let toml_file: String = if let Some(config_path) = init_match.get_one::<PathBuf>("config") {
         // load and parse the file
-        std::fs::read_to_string(std::path::Path::new(config_path))
-            .with_context(|| format!("Could not read configuration file '{config_path}'"))?
+        std::fs::read_to_string(config_path).with_context(|| {
+            format!(
+                "Could not read configuration file '{}'",
+                config_path.display()
+            )
+        })?
     } else {
         // Built-in default config
         default_conf(&EtcdBroker::locate_etcd()?)
     };
 
     let pg_version = init_match
-        .value_of("pg-version")
-        .unwrap()
-        .parse::<u32>()
+        .get_one::<u32>("pg-version")
+        .copied()
         .context("Failed to parse postgres version from the argument string")?;
 
     let mut env =
@@ -541,9 +339,10 @@ fn handle_init(init_match: &ArgMatches) -> anyhow::Result<LocalEnv> {
 
 fn pageserver_config_overrides(init_match: &ArgMatches) -> Vec<&str> {
     init_match
-        .values_of("pageserver-config-override")
+        .get_many::<String>("pageserver-config-override")
         .into_iter()
         .flatten()
+        .map(|s| s.as_str())
         .collect()
 }
 
@@ -558,7 +357,7 @@ 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
-                .values_of("config")
+                .get_many::<String>("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)?;
@@ -567,9 +366,8 @@ 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
-                .value_of("pg-version")
-                .unwrap()
-                .parse::<u32>()
+                .get_one::<u32>("pg-version")
+                .copied()
                 .context("Failed to parse postgres version from the argument string")?;
 
             let timeline_info = pageserver.timeline_create(
@@ -595,7 +393,7 @@ 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
-                .values_of("config")
+                .get_many::<String>("config")
                 .map(|vals| vals.flat_map(|c| c.split_once(':')).collect())
                 .unwrap_or_default();
 
@@ -622,13 +420,12 @@ 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
-                .value_of("branch-name")
+                .get_one::<String>("branch-name")
                 .ok_or_else(|| anyhow!("No branch name provided"))?;
 
             let pg_version = create_match
-                .value_of("pg-version")
-                .unwrap()
-                .parse::<u32>()
+                .get_one::<u32>("pg-version")
+                .copied()
                 .context("Failed to parse postgres version from the argument string")?;
 
             let timeline_info =
@@ -647,35 +444,32 @@ 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
-                .value_of("node-name")
+                .get_one::<String>("node-name")
                 .ok_or_else(|| anyhow!("No node name provided"))?;
 
             // Parse base inputs
             let base_tarfile = import_match
-                .value_of("base-tarfile")
-                .map(|s| PathBuf::from_str(s).unwrap())
-                .ok_or_else(|| anyhow!("No base-tarfile provided"))?;
+                .get_one::<PathBuf>("base-tarfile")
+                .ok_or_else(|| anyhow!("No base-tarfile provided"))?
+                .to_owned();
             let base_lsn = Lsn::from_str(
                 import_match
-                    .value_of("base-lsn")
+                    .get_one::<String>("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
-                .value_of("wal-tarfile")
-                .map(|s| PathBuf::from_str(s).unwrap());
+            let wal_tarfile = import_match.get_one::<PathBuf>("wal-tarfile").cloned();
             let end_lsn = import_match
-                .value_of("end-lsn")
+                .get_one::<String>("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
-                .value_of("pg-version")
-                .unwrap()
-                .parse::<u32>()
+                .get_one::<u32>("pg-version")
+                .copied()
                 .context("Failed to parse postgres version from the argument string")?;
 
             let mut cplane = ComputeControlPlane::load(env.clone())?;
@@ -690,10 +484,11 @@ 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
-                .value_of("branch-name")
+                .get_one::<String>("branch-name")
                 .ok_or_else(|| anyhow!("No branch name provided"))?;
             let ancestor_branch_name = branch_match
-                .value_of("ancestor-branch-name")
+                .get_one::<String>("ancestor-branch-name")
+                .map(|s| s.as_str())
                 .unwrap_or(DEFAULT_BRANCH_NAME);
             let ancestor_timeline_id = env
                 .get_branch_timeline_id(ancestor_branch_name, tenant_id)
@@ -702,8 +497,8 @@ fn handle_timeline(timeline_match: &ArgMatches, env: &mut local_env::LocalEnv) -
                 })?;
 
             let start_lsn = branch_match
-                .value_of("ancestor-start-lsn")
-                .map(Lsn::from_str)
+                .get_one::<String>("ancestor-start-lsn")
+                .map(|lsn_str| Lsn::from_str(lsn_str))
                 .transpose()
                 .context("Failed to parse ancestor start Lsn from the request")?;
             let timeline_info = pageserver.timeline_create(
@@ -804,45 +599,39 @@ fn handle_pg(pg_match: &ArgMatches, env: &local_env::LocalEnv) -> Result<()> {
         }
         "create" => {
             let branch_name = sub_args
-                .value_of("branch-name")
+                .get_one::<String>("branch-name")
+                .map(|s| s.as_str())
                 .unwrap_or(DEFAULT_BRANCH_NAME);
             let node_name = sub_args
-                .value_of("node")
-                .map(ToString::to_string)
-                .unwrap_or_else(|| format!("{}_node", branch_name));
+                .get_one::<String>("node")
+                .map(|node_name| node_name.to_string())
+                .unwrap_or_else(|| format!("{branch_name}_node"));
 
             let lsn = sub_args
-                .value_of("lsn")
-                .map(Lsn::from_str)
+                .get_one::<String>("lsn")
+                .map(|lsn_str| Lsn::from_str(lsn_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> = match sub_args.value_of("port") {
-                Some(p) => Some(p.parse()?),
-                None => None,
-            };
+            let port: Option<u16> = sub_args.get_one::<u16>("port").copied();
 
             let pg_version = sub_args
-                .value_of("pg-version")
-                .unwrap()
-                .parse::<u32>()
+                .get_one::<u32>("pg-version")
+                .copied()
                 .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> = match sub_args.value_of("port") {
-                Some(p) => Some(p.parse()?),
-                None => None,
-            };
+            let port: Option<u16> = sub_args.get_one::<u16>("port").copied();
             let node_name = sub_args
-                .value_of("node")
+                .get_one::<String>("node")
                 .ok_or_else(|| anyhow!("No node name was provided to start"))?;
 
-            let node = cplane.nodes.get(&(tenant_id, node_name.to_owned()));
+            let node = cplane.nodes.get(&(tenant_id, node_name.to_string()));
 
             let auth_token = if matches!(env.pageserver.auth_type, AuthType::NeonJWT) {
                 let claims = Claims::new(Some(tenant_id), Scope::Tenant);
@@ -853,36 +642,33 @@ 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
-                    .value_of("branch-name")
+                    .get_one::<String>("branch-name")
+                    .map(|s| s.as_str())
                     .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
-                    .value_of("lsn")
-                    .map(Lsn::from_str)
+                    .get_one::<String>("lsn")
+                    .map(|lsn_str| Lsn::from_str(lsn_str))
                     .transpose()
                     .context("Failed to parse Lsn from the request")?;
                 let pg_version = sub_args
-                    .value_of("pg-version")
-                    .unwrap()
-                    .parse::<u32>()
-                    .context("Failed to parse postgres version from the argument string")?;
+                    .get_one::<u32>("pg-version")
+                    .copied()
+                    .context("Failed to `pg-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{}) {} on timeline {} ...",
-                    pg_version, node_name, timeline_id
-                );
+                println!("Starting new postgres (v{pg_version}) {node_name} on timeline {timeline_id} ...");
 
                 let node =
                     cplane.new_node(tenant_id, node_name, timeline_id, lsn, port, pg_version)?;
@@ -891,18 +677,18 @@ fn handle_pg(pg_match: &ArgMatches, env: &local_env::LocalEnv) -> Result<()> {
         }
         "stop" => {
             let node_name = sub_args
-                .value_of("node")
+                .get_one::<String>("node")
                 .ok_or_else(|| anyhow!("No node name was provided to stop"))?;
-            let destroy = sub_args.is_present("destroy");
+            let destroy = sub_args.get_flag("destroy");
 
             let node = cplane
                 .nodes
-                .get(&(tenant_id, node_name.to_owned()))
-                .with_context(|| format!("postgres {} is not found", node_name))?;
+                .get(&(tenant_id, node_name.to_string()))
+                .with_context(|| format!("postgres {node_name} is not found"))?;
             node.stop(destroy)?;
         }
 
-        _ => bail!("Unexpected pg subcommand '{}'", sub_name),
+        _ => bail!("Unexpected pg subcommand '{sub_name}'"),
     }
 
     Ok(())
@@ -920,7 +706,10 @@ fn handle_pageserver(sub_match: &ArgMatches, env: &local_env::LocalEnv) -> Resul
         }
 
         Some(("stop", stop_match)) => {
-            let immediate = stop_match.value_of("stop-mode") == Some("immediate");
+            let immediate = stop_match
+                .get_one::<String>("stop-mode")
+                .map(|s| s.as_str())
+                == Some("immediate");
 
             if let Err(e) = pageserver.stop(immediate) {
                 eprintln!("pageserver stop failed: {}", e);
@@ -970,7 +759,7 @@ 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.value_of("id") {
+    let sk_id = if let Some(id_str) = sub_args.get_one::<String>("id") {
         NodeId(id_str.parse().context("while parsing safekeeper id")?)
     } else {
         DEFAULT_SAFEKEEPER_ID
@@ -986,7 +775,8 @@ fn handle_safekeeper(sub_match: &ArgMatches, env: &local_env::LocalEnv) -> Resul
         }
 
         "stop" => {
-            let immediate = sub_args.value_of("stop-mode") == Some("immediate");
+            let immediate =
+                sub_args.get_one::<String>("stop-mode").map(|s| s.as_str()) == Some("immediate");
 
             if let Err(e) = safekeeper.stop(immediate) {
                 eprintln!("safekeeper stop failed: {}", e);
@@ -995,7 +785,8 @@ fn handle_safekeeper(sub_match: &ArgMatches, env: &local_env::LocalEnv) -> Resul
         }
 
         "restart" => {
-            let immediate = sub_args.value_of("stop-mode") == Some("immediate");
+            let immediate =
+                sub_args.get_one::<String>("stop-mode").map(|s| s.as_str()) == Some("immediate");
 
             if let Err(e) = safekeeper.stop(immediate) {
                 eprintln!("safekeeper stop failed: {}", e);
@@ -1039,7 +830,8 @@ 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.value_of("stop-mode") == Some("immediate");
+    let immediate =
+        sub_match.get_one::<String>("stop-mode").map(|s| s.as_str()) == Some("immediate");
 
     let pageserver = PageServerNode::from_env(env);
 
@@ -1072,3 +864,219 @@ 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

@@ -12,15 +12,14 @@ use std::time::Duration;
 
 use anyhow::{Context, Result};
 use utils::{
-    connstring::connection_host_port,
     id::{TenantId, TimelineId},
     lsn::Lsn,
     postgres_backend::AuthType,
 };
 
 use crate::local_env::{LocalEnv, DEFAULT_PG_VERSION};
+use crate::pageserver::PageServerNode;
 use crate::postgresql_conf::PostgresConf;
-use crate::storage::PageServerNode;
 
 //
 // ComputeControlPlane
@@ -183,18 +182,18 @@ 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())
@@ -282,9 +281,7 @@ impl PostgresNode {
     fn setup_pg_conf(&self, auth_type: AuthType) -> Result<()> {
         let mut conf = PostgresConf::new();
         conf.append("max_wal_senders", "10");
-        // 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("wal_log_hints", "off");
         conf.append("max_replication_slots", "10");
         conf.append("hot_standby", "on");
         conf.append("shared_buffers", "1MB");
@@ -302,7 +299,8 @@ impl PostgresNode {
 
         // Configure the node to fetch pages from pageserver
         let pageserver_connstr = {
-            let (host, port) = connection_host_port(&self.pageserver.pg_connection_config);
+            let config = &self.pageserver.pg_connection_config;
+            let (host, port) = (config.host(), config.port());
 
             // Set up authentication
             //
@@ -422,7 +420,7 @@ 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(
             [
@@ -440,11 +438,11 @@ 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/connection.rs

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

control_plane/src/etcd.rs

@@ -1,95 +1,75 @@
-use std::{
-    fs,
-    path::PathBuf,
-    process::{Command, Stdio},
-};
+use std::{fs, path::PathBuf};
 
 use anyhow::Context;
-use nix::{
-    sys::signal::{kill, Signal},
-    unistd::Pid,
-};
 
-use crate::{local_env, read_pidfile};
+use crate::{background_process, local_env};
 
 pub fn start_etcd_process(env: &local_env::LocalEnv) -> anyhow::Result<()> {
     let etcd_broker = &env.etcd_broker;
     println!(
-        "Starting etcd broker using {}",
-        etcd_broker.etcd_binary_path.display()
+        "Starting etcd broker using {:?}",
+        etcd_broker.etcd_binary_path
     );
 
     let etcd_data_dir = env.base_data_dir.join("etcd");
-    fs::create_dir_all(&etcd_data_dir).with_context(|| {
-        format!(
-            "Failed to create etcd data dir: {}",
-            etcd_data_dir.display()
-        )
-    })?;
+    fs::create_dir_all(&etcd_data_dir)
+        .with_context(|| format!("Failed to create etcd data dir {etcd_data_dir:?}"))?;
 
-    let etcd_stdout_file =
-        fs::File::create(etcd_data_dir.join("etcd.stdout.log")).with_context(|| {
-            format!(
-                "Failed to create etcd stout file in directory {}",
-                etcd_data_dir.display()
-            )
-        })?;
-    let etcd_stderr_file =
-        fs::File::create(etcd_data_dir.join("etcd.stderr.log")).with_context(|| {
-            format!(
-                "Failed to create etcd stderr file in directory {}",
-                etcd_data_dir.display()
-            )
-        })?;
     let client_urls = etcd_broker.comma_separated_endpoints();
+    let args = [
+        format!("--data-dir={}", etcd_data_dir.display()),
+        format!("--listen-client-urls={client_urls}"),
+        format!("--advertise-client-urls={client_urls}"),
+        // Set --quota-backend-bytes to keep the etcd virtual memory
+        // 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(),
+    ];
 
-    let etcd_process = Command::new(&etcd_broker.etcd_binary_path)
-        .args(&[
-            format!("--data-dir={}", etcd_data_dir.display()),
-            format!("--listen-client-urls={client_urls}"),
-            format!("--advertise-client-urls={client_urls}"),
-            // Set --quota-backend-bytes to keep the etcd virtual memory
-            // size smaller. Our test etcd clusters are very small.
-            // See https://github.com/etcd-io/etcd/issues/7910
-            "--quota-backend-bytes=100000000".to_string(),
-        ])
-        .stdout(Stdio::from(etcd_stdout_file))
-        .stderr(Stdio::from(etcd_stderr_file))
-        .spawn()
-        .context("Failed to spawn etcd subprocess")?;
-    let pid = etcd_process.id();
+    let pid_file_path = etcd_pid_file_path(env);
 
-    let etcd_pid_file_path = etcd_pid_file_path(env);
-    fs::write(&etcd_pid_file_path, pid.to_string()).with_context(|| {
-        format!(
-            "Failed to create etcd pid file at {}",
-            etcd_pid_file_path.display()
-        )
-    })?;
+    let client = reqwest::blocking::Client::new();
+
+    background_process::start_process(
+        "etcd",
+        &etcd_data_dir,
+        &etcd_broker.etcd_binary_path,
+        &args,
+        background_process::InitialPidFile::Create(&pid_file_path),
+        || {
+            for broker_endpoint in &etcd_broker.broker_endpoints {
+                let request = broker_endpoint
+                    .join("health")
+                    .with_context(|| {
+                        format!(
+                            "Failed to append /health path to broker endopint {}",
+                            broker_endpoint
+                        )
+                    })
+                    .and_then(|url| {
+                        client.get(&url.to_string()).build().with_context(|| {
+                            format!("Failed to construct request to etcd endpoint {url}")
+                        })
+                    })?;
+                if client.execute(request).is_ok() {
+                    return Ok(true);
+                }
+            }
+
+            Ok(false)
+        },
+    )
+    .context("Failed to spawn etcd subprocess")?;
 
     Ok(())
 }
 
 pub fn stop_etcd_process(env: &local_env::LocalEnv) -> anyhow::Result<()> {
-    let etcd_path = &env.etcd_broker.etcd_binary_path;
-    println!("Stopping etcd broker at {}", etcd_path.display());
-
-    let etcd_pid_file_path = etcd_pid_file_path(env);
-    let pid = Pid::from_raw(read_pidfile(&etcd_pid_file_path).with_context(|| {
-        format!(
-            "Failed to read etcd pid file at {}",
-            etcd_pid_file_path.display()
-        )
-    })?);
-
-    kill(pid, Signal::SIGTERM).with_context(|| {
-        format!(
-            "Failed to stop etcd with pid {pid} at {}",
-            etcd_pid_file_path.display()
-        )
-    })?;
-
-    Ok(())
+    background_process::stop_process(true, "etcd", &etcd_pid_file_path(env))
 }
 
 fn etcd_pid_file_path(env: &local_env::LocalEnv) -> PathBuf {

control_plane/src/lib.rs

@@ -6,59 +6,12 @@
 // Intended to be used in integration tests and in CLI tools for
 // local installations.
 //
-use anyhow::{anyhow, bail, Context, Result};
-use std::fs;
-use std::path::Path;
-use std::process::Command;
 
+mod background_process;
 pub mod compute;
+pub mod connection;
 pub mod etcd;
 pub mod local_env;
+pub mod pageserver;
 pub mod postgresql_conf;
 pub mod safekeeper;
-pub mod storage;
-
-/// Read a PID file
-///
-/// We expect a file that contains a single integer.
-/// We return an i32 for compatibility with libc and nix.
-pub fn read_pidfile(pidfile: &Path) -> Result<i32> {
-    let pid_str = fs::read_to_string(pidfile)
-        .with_context(|| format!("failed to read pidfile {:?}", pidfile))?;
-    let pid: i32 = pid_str
-        .parse()
-        .map_err(|_| anyhow!("failed to parse pidfile {:?}", pidfile))?;
-    if pid < 1 {
-        bail!("pidfile {:?} contained bad value '{}'", pidfile, pid);
-    }
-    Ok(pid)
-}
-
-fn fill_rust_env_vars(cmd: &mut Command) -> &mut Command {
-    let cmd = cmd.env_clear().env("RUST_BACKTRACE", "1");
-
-    let var = "LLVM_PROFILE_FILE";
-    if let Some(val) = std::env::var_os(var) {
-        cmd.env(var, val);
-    }
-
-    const RUST_LOG_KEY: &str = "RUST_LOG";
-    if let Ok(rust_log_value) = std::env::var(RUST_LOG_KEY) {
-        cmd.env(RUST_LOG_KEY, rust_log_value)
-    } else {
-        cmd
-    }
-}
-
-fn fill_aws_secrets_vars(mut cmd: &mut Command) -> &mut Command {
-    for env_key in [
-        "AWS_ACCESS_KEY_ID",
-        "AWS_SECRET_ACCESS_KEY",
-        "AWS_SESSION_TOKEN",
-    ] {
-        if let Ok(value) = std::env::var(env_key) {
-            cmd = cmd.env(env_key, value);
-        }
-    }
-    cmd
-}

control_plane/src/local_env.rs

@@ -201,37 +201,37 @@ impl LocalEnv {
         self.pg_distrib_dir.clone()
     }
 
-    pub fn pg_distrib_dir(&self, pg_version: u32) -> PathBuf {
+    pub fn pg_distrib_dir(&self, pg_version: u32) -> anyhow::Result<PathBuf> {
         let path = self.pg_distrib_dir.clone();
 
         match pg_version {
-            14 => path.join(format!("v{pg_version}")),
-            15 => path.join(format!("v{pg_version}")),
-            _ => panic!("Unsupported postgres version: {}", pg_version),
+            14 => Ok(path.join(format!("v{pg_version}"))),
+            15 => Ok(path.join(format!("v{pg_version}"))),
+            _ => bail!("Unsupported postgres version: {}", pg_version),
         }
     }
 
-    pub fn pg_bin_dir(&self, pg_version: u32) -> PathBuf {
+    pub fn pg_bin_dir(&self, pg_version: u32) -> anyhow::Result<PathBuf> {
         match 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),
+            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),
         }
     }
-    pub fn pg_lib_dir(&self, pg_version: u32) -> PathBuf {
+    pub fn pg_lib_dir(&self, pg_version: u32) -> anyhow::Result<PathBuf> {
         match 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),
+            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),
         }
     }
 
-    pub fn pageserver_bin(&self) -> anyhow::Result<PathBuf> {
-        Ok(self.neon_distrib_dir.join("pageserver"))
+    pub fn pageserver_bin(&self) -> PathBuf {
+        self.neon_distrib_dir.join("pageserver")
     }
 
-    pub fn safekeeper_bin(&self) -> anyhow::Result<PathBuf> {
-        Ok(self.neon_distrib_dir.join("safekeeper"))
+    pub fn safekeeper_bin(&self) -> PathBuf {
+        self.neon_distrib_dir.join("safekeeper")
     }
 
     pub fn pg_data_dirs_path(&self) -> PathBuf {
@@ -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/pageserver.rs

@@ -1,33 +1,27 @@
 use std::collections::HashMap;
-use std::fs::File;
+use std::fs::{self, File};
 use std::io::{BufReader, Write};
 use std::num::NonZeroU64;
 use std::path::{Path, PathBuf};
-use std::process::Command;
-use std::time::Duration;
-use std::{io, result, thread};
+use std::process::Child;
+use std::{io, result};
 
+use crate::connection::PgConnectionConfig;
 use anyhow::{bail, Context};
-use nix::errno::Errno;
-use nix::sys::signal::{kill, Signal};
-use nix::unistd::Pid;
 use pageserver_api::models::{
     TenantConfigRequest, TenantCreateRequest, TenantInfo, TimelineCreateRequest, TimelineInfo,
 };
-use postgres::{Config, NoTls};
 use reqwest::blocking::{Client, RequestBuilder, Response};
 use reqwest::{IntoUrl, Method};
 use thiserror::Error;
 use utils::{
-    connstring::connection_address,
     http::error::HttpErrorBody,
     id::{TenantId, TimelineId},
     lsn::Lsn,
     postgres_backend::AuthType,
 };
 
-use crate::local_env::LocalEnv;
-use crate::{fill_aws_secrets_vars, fill_rust_env_vars, read_pidfile};
+use crate::{background_process, local_env::LocalEnv};
 
 #[derive(Error, Debug)]
 pub enum PageserverHttpError {
@@ -75,7 +69,7 @@ impl ResponseErrorMessageExt for Response {
 //
 #[derive(Debug)]
 pub struct PageServerNode {
-    pub pg_connection_config: Config,
+    pub pg_connection_config: PgConnectionConfig,
     pub env: LocalEnv,
     pub http_client: Client,
     pub http_base_url: String,
@@ -101,7 +95,7 @@ impl PageServerNode {
     }
 
     /// Construct libpq connection string for connecting to the pageserver.
-    fn pageserver_connection_config(password: &str, listen_addr: &str) -> Config {
+    fn pageserver_connection_config(password: &str, listen_addr: &str) -> PgConnectionConfig {
         format!("postgresql://no_user:{password}@{listen_addr}/no_db")
             .parse()
             .unwrap()
@@ -161,7 +155,15 @@ impl PageServerNode {
             init_config_overrides.push("auth_validation_public_key_path='auth_public_key.pem'");
         }
 
-        self.start_node(&init_config_overrides, &self.env.base_data_dir, true)?;
+        let mut pageserver_process = self
+            .start_node(&init_config_overrides, &self.env.base_data_dir, true)
+            .with_context(|| {
+                format!(
+                    "Failed to start a process for pageserver {}",
+                    self.env.pageserver.id,
+                )
+            })?;
+
         let init_result = self
             .try_init_timeline(create_tenant, initial_timeline_id, pg_version)
             .context("Failed to create initial tenant and timeline for pageserver");
@@ -171,7 +173,29 @@ impl PageServerNode {
             }
             Err(e) => eprintln!("{e:#}"),
         }
-        self.stop(false)?;
+        match pageserver_process.kill() {
+            Err(e) => {
+                eprintln!(
+                    "Failed to stop pageserver {} process with pid {}: {e:#}",
+                    self.env.pageserver.id,
+                    pageserver_process.id(),
+                )
+            }
+            Ok(()) => {
+                println!(
+                    "Stopped pageserver {} process with pid {}",
+                    self.env.pageserver.id,
+                    pageserver_process.id(),
+                );
+                // cleanup after pageserver startup, since we do not call regular `stop_process` during init
+                let pid_file = self.pid_file();
+                if let Err(e) = fs::remove_file(&pid_file) {
+                    if e.kind() != io::ErrorKind::NotFound {
+                        eprintln!("Failed to remove pid file {pid_file:?} after stopping the process: {e:#}");
+                    }
+                }
+            }
+        }
         init_result
     }
 
@@ -196,11 +220,14 @@ impl PageServerNode {
         self.env.pageserver_data_dir()
     }
 
-    pub fn pid_file(&self) -> PathBuf {
+    /// The pid file is created by the pageserver process, with its pid stored inside.
+    /// Other pageservers cannot lock the same file and overwrite it for as long as the current
+    /// pageserver runs. (Unless someone removes the file manually; never do that!)
+    fn pid_file(&self) -> PathBuf {
         self.repo_path().join("pageserver.pid")
     }
 
-    pub fn start(&self, config_overrides: &[&str]) -> anyhow::Result<()> {
+    pub fn start(&self, config_overrides: &[&str]) -> anyhow::Result<Child> {
         self.start_node(config_overrides, &self.repo_path(), false)
     }
 
@@ -209,10 +236,10 @@ impl PageServerNode {
         config_overrides: &[&str],
         datadir: &Path,
         update_config: bool,
-    ) -> anyhow::Result<()> {
+    ) -> anyhow::Result<Child> {
         println!(
             "Starting pageserver at '{}' in '{}'",
-            connection_address(&self.pg_connection_config),
+            self.pg_connection_config.raw_address(),
             datadir.display()
         );
         io::stdout().flush()?;
@@ -220,10 +247,7 @@ impl PageServerNode {
         let mut args = vec![
             "-D",
             datadir.to_str().with_context(|| {
-                format!(
-                    "Datadir path '{}' cannot be represented as a unicode string",
-                    datadir.display()
-                )
+                format!("Datadir path {datadir:?} cannot be represented as a unicode string")
             })?,
         ];
 
@@ -235,48 +259,18 @@ impl PageServerNode {
             args.extend(["-c", config_override]);
         }
 
-        let mut cmd = Command::new(self.env.pageserver_bin()?);
-        let mut filled_cmd = fill_rust_env_vars(cmd.args(&args).arg("--daemonize"));
-        filled_cmd = fill_aws_secrets_vars(filled_cmd);
-
-        if !filled_cmd.status()?.success() {
-            bail!(
-                "Pageserver failed to start. See console output and '{}' for details.",
-                datadir.join("pageserver.log").display()
-            );
-        }
-
-        // It takes a while for the page server to start up. Wait until it is
-        // open for business.
-        const RETRIES: i8 = 15;
-        for retries in 1..RETRIES {
-            match self.check_status() {
-                Ok(()) => {
-                    println!("\nPageserver started");
-                    return Ok(());
-                }
-                Err(err) => {
-                    match err {
-                        PageserverHttpError::Transport(err) => {
-                            if err.is_connect() && retries < 5 {
-                                print!(".");
-                                io::stdout().flush().unwrap();
-                            } else {
-                                if retries == 5 {
-                                    println!() // put a line break after dots for second message
-                                }
-                                println!("Pageserver not responding yet, err {err} retrying ({retries})...");
-                            }
-                        }
-                        PageserverHttpError::Response(msg) => {
-                            bail!("pageserver failed to start: {msg} ")
-                        }
-                    }
-                    thread::sleep(Duration::from_secs(1));
-                }
-            }
-        }
-        bail!("pageserver failed to start in {RETRIES} seconds");
+        background_process::start_process(
+            "pageserver",
+            datadir,
+            &self.env.pageserver_bin(),
+            &args,
+            background_process::InitialPidFile::Expect(&self.pid_file()),
+            || match self.check_status() {
+                Ok(()) => Ok(true),
+                Err(PageserverHttpError::Transport(_)) => Ok(false),
+                Err(e) => Err(anyhow::anyhow!("Failed to check node status: {e}")),
+            },
+        )
     }
 
     ///
@@ -288,69 +282,18 @@ impl PageServerNode {
     /// If the server is not running, returns success
     ///
     pub fn stop(&self, immediate: bool) -> anyhow::Result<()> {
-        let pid_file = self.pid_file();
-        if !pid_file.exists() {
-            println!("Pageserver is already stopped");
-            return Ok(());
-        }
-        let pid = Pid::from_raw(read_pidfile(&pid_file)?);
-
-        let sig = if immediate {
-            print!("Stopping pageserver immediately..");
-            Signal::SIGQUIT
-        } else {
-            print!("Stopping pageserver gracefully..");
-            Signal::SIGTERM
-        };
-        io::stdout().flush().unwrap();
-        match kill(pid, sig) {
-            Ok(_) => (),
-            Err(Errno::ESRCH) => {
-                println!("Pageserver with pid {pid} does not exist, but a PID file was found");
-                return Ok(());
-            }
-            Err(err) => bail!(
-                "Failed to send signal to pageserver with pid {pid}: {}",
-                err.desc()
-            ),
-        }
-
-        // Wait until process is gone
-        for i in 0..600 {
-            let signal = None; // Send no signal, just get the error code
-            match kill(pid, signal) {
-                Ok(_) => (), // Process exists, keep waiting
-                Err(Errno::ESRCH) => {
-                    // Process not found, we're done
-                    println!("done!");
-                    return Ok(());
-                }
-                Err(err) => bail!(
-                    "Failed to send signal to pageserver with pid {}: {}",
-                    pid,
-                    err.desc()
-                ),
-            };
-
-            if i % 10 == 0 {
-                print!(".");
-                io::stdout().flush().unwrap();
-            }
-            thread::sleep(Duration::from_millis(100));
-        }
-
-        bail!("Failed to stop pageserver with pid {pid}");
+        background_process::stop_process(immediate, "pageserver", &self.pid_file())
     }
 
     pub fn page_server_psql(&self, sql: &str) -> Vec<postgres::SimpleQueryMessage> {
-        let mut client = self.pg_connection_config.connect(NoTls).unwrap();
+        let mut client = self.pg_connection_config.connect_no_tls().unwrap();
 
         println!("Pageserver query: '{sql}'");
         client.simple_query(sql).unwrap()
     }
 
     pub fn page_server_psql_client(&self) -> result::Result<postgres::Client, postgres::Error> {
-        self.pg_connection_config.connect(NoTls)
+        self.pg_connection_config.connect_no_tls()
     }
 
     fn http_request<U: IntoUrl>(&self, method: Method, url: U) -> RequestBuilder {
@@ -419,6 +362,11 @@ impl PageServerNode {
                 .map(|x| x.parse::<NonZeroU64>())
                 .transpose()
                 .context("Failed to parse 'max_lsn_wal_lag' as non zero integer")?,
+            trace_read_requests: settings
+                .remove("trace_read_requests")
+                .map(|x| x.parse::<bool>())
+                .transpose()
+                .context("Failed to parse 'trace_read_requests' as bool")?,
         };
         if !settings.is_empty() {
             bail!("Unrecognized tenant settings: {settings:?}")
@@ -481,6 +429,11 @@ impl PageServerNode {
                     .map(|x| x.parse::<NonZeroU64>())
                     .transpose()
                     .context("Failed to parse 'max_lsn_wal_lag' as non zero integer")?,
+                trace_read_requests: settings
+                    .get("trace_read_requests")
+                    .map(|x| x.parse::<bool>())
+                    .transpose()
+                    .context("Failed to parse 'trace_read_requests' as bool")?,
             })
             .send()?
             .error_from_body()?;
@@ -549,7 +502,7 @@ impl PageServerNode {
         pg_wal: Option<(Lsn, PathBuf)>,
         pg_version: u32,
     ) -> anyhow::Result<()> {
-        let mut client = self.pg_connection_config.connect(NoTls).unwrap();
+        let mut client = self.pg_connection_config.connect_no_tls().unwrap();
 
         // Init base reader
         let (start_lsn, base_tarfile_path) = base;

control_plane/src/safekeeper.rs

@@ -1,23 +1,21 @@
 use std::io::Write;
 use std::path::PathBuf;
-use std::process::Command;
+use std::process::Child;
 use std::sync::Arc;
-use std::time::Duration;
-use std::{io, result, thread};
+use std::{io, result};
 
-use anyhow::bail;
-use nix::errno::Errno;
-use nix::sys::signal::{kill, Signal};
-use nix::unistd::Pid;
-use postgres::Config;
+use anyhow::Context;
 use reqwest::blocking::{Client, RequestBuilder, Response};
 use reqwest::{IntoUrl, Method};
 use thiserror::Error;
-use utils::{connstring::connection_address, http::error::HttpErrorBody, id::NodeId};
+use utils::{http::error::HttpErrorBody, id::NodeId};
 
-use crate::local_env::{LocalEnv, SafekeeperConf};
-use crate::storage::PageServerNode;
-use crate::{fill_aws_secrets_vars, fill_rust_env_vars, read_pidfile};
+use crate::connection::PgConnectionConfig;
+use crate::pageserver::PageServerNode;
+use crate::{
+    background_process,
+    local_env::{LocalEnv, SafekeeperConf},
+};
 
 #[derive(Error, Debug)]
 pub enum SafekeeperHttpError {
@@ -63,7 +61,7 @@ pub struct SafekeeperNode {
 
     pub conf: SafekeeperConf,
 
-    pub pg_connection_config: Config,
+    pub pg_connection_config: PgConnectionConfig,
     pub env: LocalEnv,
     pub http_client: Client,
     pub http_base_url: String,
@@ -87,15 +85,15 @@ impl SafekeeperNode {
     }
 
     /// Construct libpq connection string for connecting to this safekeeper.
-    fn safekeeper_connection_config(port: u16) -> Config {
+    fn safekeeper_connection_config(port: u16) -> PgConnectionConfig {
         // TODO safekeeper authentication not implemented yet
-        format!("postgresql://no_user@127.0.0.1:{}/no_db", port)
+        format!("postgresql://no_user@127.0.0.1:{port}/no_db")
             .parse()
             .unwrap()
     }
 
     pub fn datadir_path_by_id(env: &LocalEnv, sk_id: NodeId) -> PathBuf {
-        env.safekeeper_data_dir(format!("sk{}", sk_id).as_ref())
+        env.safekeeper_data_dir(&format!("sk{sk_id}"))
     }
 
     pub fn datadir_path(&self) -> PathBuf {
@@ -106,92 +104,78 @@ impl SafekeeperNode {
         self.datadir_path().join("safekeeper.pid")
     }
 
-    pub fn start(&self) -> anyhow::Result<()> {
+    pub fn start(&self) -> anyhow::Result<Child> {
         print!(
             "Starting safekeeper at '{}' in '{}'",
-            connection_address(&self.pg_connection_config),
+            self.pg_connection_config.raw_address(),
             self.datadir_path().display()
         );
         io::stdout().flush().unwrap();
 
         let listen_pg = format!("127.0.0.1:{}", self.conf.pg_port);
         let listen_http = format!("127.0.0.1:{}", self.conf.http_port);
+        let id = self.id;
+        let datadir = self.datadir_path();
 
-        let mut cmd = Command::new(self.env.safekeeper_bin()?);
-        fill_rust_env_vars(
-            cmd.args(&["-D", self.datadir_path().to_str().unwrap()])
-                .args(&["--id", self.id.to_string().as_ref()])
-                .args(&["--listen-pg", &listen_pg])
-                .args(&["--listen-http", &listen_http])
-                .args(&["--recall", "1 second"])
-                .arg("--daemonize"),
-        );
+        let id_string = id.to_string();
+        let mut args = vec![
+            "-D",
+            datadir.to_str().with_context(|| {
+                format!("Datadir path {datadir:?} cannot be represented as a unicode string")
+            })?,
+            "--id",
+            &id_string,
+            "--listen-pg",
+            &listen_pg,
+            "--listen-http",
+            &listen_http,
+        ];
         if !self.conf.sync {
-            cmd.arg("--no-sync");
+            args.push("--no-sync");
         }
 
         let comma_separated_endpoints = self.env.etcd_broker.comma_separated_endpoints();
         if !comma_separated_endpoints.is_empty() {
-            cmd.args(&["--broker-endpoints", &comma_separated_endpoints]);
+            args.extend(["--broker-endpoints", &comma_separated_endpoints]);
         }
         if let Some(prefix) = self.env.etcd_broker.broker_etcd_prefix.as_deref() {
-            cmd.args(&["--broker-etcd-prefix", prefix]);
+            args.extend(["--broker-etcd-prefix", prefix]);
         }
+
+        let mut backup_threads = String::new();
         if let Some(threads) = self.conf.backup_threads {
-            cmd.args(&["--backup-threads", threads.to_string().as_ref()]);
+            backup_threads = threads.to_string();
+            args.extend(["--backup-threads", &backup_threads]);
+        } else {
+            drop(backup_threads);
         }
+
         if let Some(ref remote_storage) = self.conf.remote_storage {
-            cmd.args(&["--remote-storage", remote_storage]);
+            args.extend(["--remote-storage", remote_storage]);
         }
+
+        let key_path = self.env.base_data_dir.join("auth_public_key.pem");
         if self.conf.auth_enabled {
-            cmd.arg("--auth-validation-public-key-path");
-            // PathBuf is better be passed as is, not via `String`.
-            cmd.arg(self.env.base_data_dir.join("auth_public_key.pem"));
+            args.extend([
+                "--auth-validation-public-key-path",
+                key_path.to_str().with_context(|| {
+                    format!("Key path {key_path:?} cannot be represented as a unicode string")
+                })?,
+            ]);
         }
 
-        fill_aws_secrets_vars(&mut cmd);
-
-        if !cmd.status()?.success() {
-            bail!(
-                "Safekeeper failed to start. See '{}' for details.",
-                self.datadir_path().join("safekeeper.log").display()
-            );
-        }
-
-        // It takes a while for the safekeeper to start up. Wait until it is
-        // open for business.
-        const RETRIES: i8 = 15;
-        for retries in 1..RETRIES {
-            match self.check_status() {
-                Ok(_) => {
-                    println!("\nSafekeeper started");
-                    return Ok(());
-                }
-                Err(err) => {
-                    match err {
-                        SafekeeperHttpError::Transport(err) => {
-                            if err.is_connect() && retries < 5 {
-                                print!(".");
-                                io::stdout().flush().unwrap();
-                            } else {
-                                if retries == 5 {
-                                    println!() // put a line break after dots for second message
-                                }
-                                println!(
-                                    "Safekeeper not responding yet, err {} retrying ({})...",
-                                    err, retries
-                                );
-                            }
-                        }
-                        SafekeeperHttpError::Response(msg) => {
-                            bail!("safekeeper failed to start: {} ", msg)
-                        }
-                    }
-                    thread::sleep(Duration::from_secs(1));
-                }
-            }
-        }
-        bail!("safekeeper failed to start in {} seconds", RETRIES);
+        background_process::start_process(
+            &format!("safekeeper {id}"),
+            &datadir,
+            &self.env.safekeeper_bin(),
+            &args,
+            background_process::InitialPidFile::Expect(&self.pid_file()),
+            || match self.check_status() {
+                Ok(()) => Ok(true),
+                Err(SafekeeperHttpError::Transport(_)) => Ok(false),
+                Err(e) => Err(anyhow::anyhow!("Failed to check node status: {e}")),
+            },
+        )
     }
 
     ///
@@ -203,63 +187,11 @@ impl SafekeeperNode {
     /// If the server is not running, returns success
     ///
     pub fn stop(&self, immediate: bool) -> anyhow::Result<()> {
-        let pid_file = self.pid_file();
-        if !pid_file.exists() {
-            println!("Safekeeper {} is already stopped", self.id);
-            return Ok(());
-        }
-        let pid = read_pidfile(&pid_file)?;
-        let pid = Pid::from_raw(pid);
-
-        let sig = if immediate {
-            print!("Stopping safekeeper {} immediately..", self.id);
-            Signal::SIGQUIT
-        } else {
-            print!("Stopping safekeeper {} gracefully..", self.id);
-            Signal::SIGTERM
-        };
-        io::stdout().flush().unwrap();
-        match kill(pid, sig) {
-            Ok(_) => (),
-            Err(Errno::ESRCH) => {
-                println!(
-                    "Safekeeper with pid {} does not exist, but a PID file was found",
-                    pid
-                );
-                return Ok(());
-            }
-            Err(err) => bail!(
-                "Failed to send signal to safekeeper with pid {}: {}",
-                pid,
-                err.desc()
-            ),
-        }
-
-        // Wait until process is gone
-        for i in 0..600 {
-            let signal = None; // Send no signal, just get the error code
-            match kill(pid, signal) {
-                Ok(_) => (), // Process exists, keep waiting
-                Err(Errno::ESRCH) => {
-                    // Process not found, we're done
-                    println!("done!");
-                    return Ok(());
-                }
-                Err(err) => bail!(
-                    "Failed to send signal to pageserver with pid {}: {}",
-                    pid,
-                    err.desc()
-                ),
-            };
-
-            if i % 10 == 0 {
-                print!(".");
-                io::stdout().flush().unwrap();
-            }
-            thread::sleep(Duration::from_millis(100));
-        }
-
-        bail!("Failed to stop safekeeper with pid {}", pid);
+        background_process::stop_process(
+            immediate,
+            &format!("safekeeper {}", self.id),
+            &self.pid_file(),
+        )
     }
 
     fn http_request<U: IntoUrl>(&self, method: Method, url: U) -> RequestBuilder {

docker-compose/compute_wrapper/Dockerfile

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

docker-compose/compute_wrapper/shell/compute.sh

@@ -0,0 +1,48 @@
+#!/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_wrapper/var/db/postgres/specs/spec.json

@@ -0,0 +1,141 @@
+{
+    "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

@@ -0,0 +1,209 @@
+version: '3'
+
+services:
+  etcd:
+    restart: always
+    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:
+    restart: always
+    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:
+    restart: always
+    image: ${REPOSITORY:-neondatabase}/neon:${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=[$$BROKER_ENDPOINT]\"
+                                   -c \"listen_pg_addr='0.0.0.0:6400'\"
+                                   -c \"listen_http_addr='0.0.0.0:9898'\"
+                                   -c \"remote_storage={endpoint='http://minio:9000',
+                                                        bucket_name='neon',
+                                                        bucket_region='eu-north-1',
+                                                        prefix_in_bucket='/pageserver/'}\""
+    depends_on:
+      - etcd
+      - minio_create_buckets
+
+  safekeeper1:
+    restart: always
+    image: ${REPOSITORY:-neondatabase}/neon:${TAG:-latest}
+    environment:
+      - SAFEKEEPER_ADVERTISE_URL=safekeeper1:5454
+      - SAFEKEEPER_ID=1
+      - BROKER_ENDPOINT=http://etcd:2379
+      - AWS_ACCESS_KEY_ID=minio
+      - AWS_SECRET_ACCESS_KEY=password
+      #- RUST_BACKTRACE=1
+    ports:
+      #- 5454:5454 # pg protocol handler
+      - 7676:7676 # http endpoints
+    entrypoint:
+      - "/bin/sh"
+      - "-c"
+    command:
+      - "safekeeper --listen-pg=$$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:
+    restart: always
+    image: ${REPOSITORY:-neondatabase}/neon:${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=$$BROKER_ENDPOINT
+                    -D /data
+                    --remote-storage=\"{endpoint='http://minio:9000',
+                                        bucket_name='neon',
+                                        bucket_region='eu-north-1',
+                                        prefix_in_bucket='/safekeeper/'}\""
+    depends_on:
+      - etcd
+      - minio_create_buckets
+
+  safekeeper3:
+    restart: always
+    image: ${REPOSITORY:-neondatabase}/neon:${TAG:-latest}
+    environment:
+      - SAFEKEEPER_ADVERTISE_URL=safekeeper3:5454
+      - SAFEKEEPER_ID=3
+      - BROKER_ENDPOINT=http://etcd:2379
+      - AWS_ACCESS_KEY_ID=minio
+      - AWS_SECRET_ACCESS_KEY=password
+      #- RUST_BACKTRACE=1
+    ports:
+      #- 5454:5454 # pg protocol handler
+      - 7678:7676 # http endpoints
+    entrypoint:
+      - "/bin/sh"
+      - "-c"
+    command:
+      - "safekeeper --listen-pg=$$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:
+    restart: always
+    build:
+      context: ./compute_wrapper/
+      args:
+        - COMPUTE_IMAGE=compute-node-v${PG_VERSION:-14}
+        - TAG=${TAG:-latest}
+        - http_proxy=$http_proxy
+        - https_proxy=$https_proxy
+    environment:
+      - PG_VERSION=${PG_VERSION:-14}
+      #- RUST_BACKTRACE=1
+    # Mount the test files directly, for faster editing cycle.
+    volumes:
+      - ./compute_wrapper/var/db/postgres/specs/:/var/db/postgres/specs/
+      - ./compute_wrapper/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/docker_compose_test.sh

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

docs/SUMMARY.md

@@ -80,4 +80,6 @@
 - [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/broker.md

@@ -0,0 +1,25 @@
+# Neon broker
+
+Neon storage broker targets two issues:
+- Allowing safekeepers and pageservers learn which nodes also hold their
+  timelines, and timeline statuses there.
+- Avoiding O(n^2) connections between storage nodes while doing so.
+
+This is used
+- By pageservers to determine the most advanced and alive safekeeper to pull WAL from.
+- By safekeepers to synchronize on the timeline: advance
+  `remote_consistent_lsn`, `backup_lsn`, choose who offloads WAL to s3.
+
+Technically, it is a simple stateless pub-sub message broker based on tonic
+(grpc) making multiplexing easy. Since it is stateless, fault tolerance can be
+provided by k8s; there is no built in replication support, though it is not hard
+to add.
+
+Currently, the only message is `SafekeeperTimelineInfo`. Each safekeeper, for
+each active timeline, once in a while pushes timeline status to the broker.
+Other nodes subscribe and receive this info, using it per above.
+
+grpcurl can be used to check which values are currently being pushed:
+```
+grpcurl -proto broker/proto/broker.proto -d '{"all":{}}' -plaintext localhost:50051 neon_broker.NeonBroker/SubscribeSafekeeperInfo
+```

docs/docker.md

@@ -18,3 +18,67 @@ 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

@@ -0,0 +1,91 @@
+# 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.

docs/rfcs/020-pageserver-s3-coordination.md

@@ -0,0 +1,246 @@
+# Coordinating access of multiple pageservers to the same s3 data
+
+## Motivation
+
+There are some blind spots around coordinating access of multiple pageservers
+to the same s3 data. Currently this is applicable only to tenant relocation
+case, but in the future we'll need to solve similar problems for
+replica/standby pageservers.
+
+## Impacted components (e.g. pageserver, safekeeper, console, etc)
+
+Pageserver
+
+## The problem
+
+### Relocation
+
+During relocation both pageservers can write to s3. This should be ok for all
+data except the `index_part.json`. For index part it causes problems during
+compaction/gc because they remove files from index/s3.
+
+Imagine this case:
+
+```mermaid
+sequenceDiagram
+    autonumber
+    participant PS1
+    participant S3
+    participant PS2
+
+    PS1->>S3: Uploads L1, L2 <br/> Index contains L1 L2
+    PS2->>S3: Attach called, sees L1, L2
+    PS1->>S3: Compaction comes <br/> Removes L1, adds L3
+    note over S3: Index now L2, L3
+    PS2->>S3: Uploads new layer L4 <br/> (added to previous view of the index)
+    note over S3: Index now L1, L2, L4
+```
+
+At this point it is not possible to restore from index, it contains L2 which
+is no longer available in s3 and doesnt contain L3 added by compaction by the
+first pageserver. So if any of the pageservers restart initial sync will fail
+(or in on-demand world it will fail a bit later during page request from
+missing layer)
+
+### Standby pageserver
+
+Another related case is standby pageserver. In this case second pageserver can
+be used as a replica to scale reads and serve as a failover target in case
+first one fails.
+
+In this mode second pageserver needs to have the same picture of s3 files to
+be able to load layers on-demand. To accomplish that second pageserver
+cannot run gc/compaction jobs. Instead it needs to receive updates for index
+contents. (There is no need to run walreceiver on the second pageserver then).
+
+## Observations
+
+- If both pageservers ingest wal then their layer set diverges, because layer
+  file generation is not deterministic
+- If one of the pageservers does not ingest wal (and just picks up layer
+  updates) then it lags behind and cannot really answer queries in the same
+  pace as the primary one
+- Can compaction help make layers deterministic? E g we do not upload level
+  zero layers and construction of higher levels should be deterministic.
+  This way we can guarantee that layer creation by timeout wont mess things up.
+  This way one pageserver uploads data and second one can just ingest it.
+  But we still need some form of election
+
+## Solutions
+
+### Manual orchestration
+
+One possible solution for relocation case is to orchestrate background jobs
+from outside. The oracle who runs migration can turn off background jobs on
+PS1 before migration and then run migration -> enable them on PS2. The problem
+comes if migration fails. In this case in order to resume background jobs
+oracle needs to guarantee that PS2 doesnt run background jobs and if it doesnt
+respond then PS1 is stuck unable to run compaction/gc. This cannot be solved
+without human ensuring that no upload from PS2 can happen. In order to be able
+to resolve this automatically CAS is required on S3 side so pageserver can
+avoid overwriting index part if it is no longer the leading one
+
+Note that flag that disables background jobs needs to be persistent, because
+otherwise pageserver restart will clean it
+
+### Avoid index_part.json
+
+Index part consists of two parts, list of layers and metadata. List of layers
+can be easily obtained by `ListObjects` S3 API method. But what to do with
+metadata? Create metadata instance for each checkpoint and add some counter
+to the file name?
+
+Back to potentially long s3 ls.
+
+### Coordination based approach
+
+Do it like safekeepers chose leader for WAL upload. Ping each other and decide
+based on some heuristics e g smallest node id. During relocation PS1 sends
+"resign" ping message so others can start election without waiting for a timeout.
+
+This still leaves metadata question open and non deterministic layers are a
+problem as well
+
+### Avoid metadata file
+
+One way to eliminate metadata file is to store it in layer files under some
+special key. This may resonate with intention to keep all relation sizes in
+some special segment to avoid initial download during size calculation.
+Maybe with that we can even store pre calculated value.
+
+As a downside each checkpoint gets 512 bytes larger.
+
+If we entirely avoid metadata file this opens up many approaches
+
+* * *
+
+During discussion it seems that we converged on the approach consisting of:
+
+- index files stored per pageserver in the same timeline directory. With that
+  index file name starts to look like: `<pageserver_node_id>_index_part.json`.
+  In such set up there are no concurrent overwrites of index file by different
+  pageservers.
+- For replica pageservers the solution would be for primary to broadcast index
+  changes to any followers with an ability to check index files in s3 and
+  restore the full state. To properly merge changes with index files we can use
+  a counter that is persisted in an index file, is incremented on every change
+  to it and passed along with broadcasted change. This way we can determine
+  whether we need to apply change to the index state or not.
+- Responsibility for running background jobs is assigned externally. Pageserver
+  keeps locally persistent flag for each tenant that indicates whether this
+  pageserver is considered as primary one or not. TODO what happends if we
+  crash and cannot start for some extended period of time? Control plane can
+  assign ownership to some other pageserver. Pageserver needs some way to check
+  if its still the blessed one. Maybe by explicit request to control plane on
+  start.
+
+Requirement for deterministic layer generation was considered overly strict
+because of two reasons:
+
+- It can limit possible optimizations e g when pageserver wants to reshuffle
+  some data locally and doesnt want to coordinate this
+- The deterministic algorithm itself can change so during deployments for some
+  time there will be two different version running at the same time which can
+  cause non determinism
+
+### External elections
+
+The above case with lost state in this schema with externally managed
+leadership is represented like this:
+
+Note that here we keep objects list in the index file.
+
+```mermaid
+sequenceDiagram
+    autonumber
+    participant PS1
+    participant CP as Control Plane
+    participant S3
+    participant PS2
+
+    note over PS1,PS2: PS1 starts up and still a leader
+    PS1->>CP: Am I still the leader for Tenant X?
+    activate CP
+    CP->>PS1: Yes
+    deactivate CP
+    PS1->>S3: Fetch PS1 index.
+    note over PS1: Continue operations, start backround jobs
+    note over PS1,PS2: PS1 starts up and still and is not a leader anymore
+    PS1->>CP: Am I still the leader for Tenant X?
+    CP->>PS1: No
+    PS1->>PS2: Subscribe to index changes
+    PS1->>S3: Fetch PS1 and PS2 indexes
+    note over PS1: Combine index file to include layers <br> from both indexes to be able <br> to see newer files from leader (PS2)
+    note over PS1: Continue operations, do not start background jobs
+```
+
+### Internal elections
+
+To manage leadership internally we can use broker to exchange pings so nodes
+can decide on the leader roles. In case multiple pageservers are active leader
+is the one with lowest node id.
+
+Operations with internally managed elections:
+
+```mermaid
+sequenceDiagram
+    autonumber
+    participant PS1
+    participant S3
+
+    note over PS1: Starts up
+    note over PS1: Subscribes to changes, waits for two ping <br> timeouts to see if there is a leader
+    PS1->>S3: Fetch indexes from s3
+    alt there is a leader
+        note over PS1: do not start background jobs, <br> continue applying index updates
+    else there is no leader
+        note over PS1: start background jobs, <br> broadcast index changes
+    end
+
+    note over PS1,S3: Then the picture is similar to external elections <br> the difference is that follower can become a leader <br> if there are no pings after some timeout new leader gets elected
+```
+
+### Eviction
+
+When two pageservers operate on a tenant for extended period of time follower
+doesnt perform write operations in s3. When layer is evicted follower relies
+on updates from primary to get info about layers it needs to cover range for
+evicted layer.
+
+Note that it wont match evicted layer exactly, so layers will overlap and
+lookup code needs to correctly handle that.
+
+### Relocation flow
+
+Actions become:
+
+- Attach tenant to new pageserver
+- New pageserver becomes follower since previous one is still leading
+- New pageserver starts replicating from safekeepers but does not upload layers
+- Detach is called on the old one
+- New pageserver becomes leader after it realizes that old one disappeared
+
+### Index File
+
+Using `s3 ls` on startup simplifies things, but we still need metadata, so we
+need to fetch index files anyway. If they contain list of files we can combine
+them and avoid costly `s3 ls`
+
+### Remaining issues
+
+- More than one remote consistent lsn for safekeepers to know
+
+Anything else?
+
+### Proposed solution
+
+To recap. On meeting we converged on approach with external elections but I
+think it will be overall harder to manage and will introduce a dependency on
+control plane for pageserver. Using separate index files for each pageserver
+consisting of log of operations and a metadata snapshot should be enough.
+
+### What we need to get there?
+
+- Change index file structure to contain log of changes instead of just the
+  file list
+- Implement pinging/elections for pageservers

docs/sourcetree.md

@@ -2,6 +2,11 @@
 
 Below you will find a brief overview of each subdir in the source tree in alphabetical order.
 
+`broker`:
+
+Neon storage broker, providing messaging between safekeepers and pageservers.
+[broker.md](./broker.md)
+
 `/control_plane`:
 
 Local control plane.
@@ -52,6 +57,10 @@ PostgreSQL extension that implements storage manager API and network communicati
 
 PostgreSQL extension that contains functions needed for testing and debugging.
 
+`/pgxn/neon_walredo`:
+
+Library to run Postgres as a "WAL redo process" in the pageserver.
+
 `/safekeeper`:
 
 The neon WAL service that receives WAL from a primary compute nodes and streams it to the pageserver.

libs/pageserver_api/Cargo.toml

@@ -7,6 +7,10 @@ edition = "2021"
 serde = { version = "1.0", features = ["derive"] }
 serde_with = "2.0"
 const_format = "0.2.21"
+anyhow = { version = "1.0", features = ["backtrace"] }
+bytes = "1.0.1"
+byteorder = "1.4.3"
 
 utils = { path = "../utils" }
+postgres_ffi = { path = "../postgres_ffi" }
 workspace_hack = { version = "0.1", path = "../../workspace_hack" }

libs/pageserver_api/src/lib.rs

@@ -2,6 +2,7 @@ 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

@@ -1,5 +1,6 @@
 use std::num::NonZeroU64;
 
+use byteorder::{BigEndian, ReadBytesExt};
 use serde::{Deserialize, Serialize};
 use serde_with::{serde_as, DisplayFromStr};
 use utils::{
@@ -7,6 +8,10 @@ use utils::{
     lsn::Lsn,
 };
 
+use crate::reltag::RelTag;
+use anyhow::bail;
+use bytes::{BufMut, Bytes, BytesMut};
+
 /// A state of a tenant in pageserver's memory.
 #[derive(Debug, Clone, Copy, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
 pub enum TenantState {
@@ -19,6 +24,22 @@ 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 {
@@ -52,6 +73,7 @@ pub struct TenantCreateRequest {
     pub walreceiver_connect_timeout: Option<String>,
     pub lagging_wal_timeout: Option<String>,
     pub max_lsn_wal_lag: Option<NonZeroU64>,
+    pub trace_read_requests: Option<bool>,
 }
 
 #[serde_as]
@@ -91,6 +113,7 @@ pub struct TenantConfigRequest {
     pub walreceiver_connect_timeout: Option<String>,
     pub lagging_wal_timeout: Option<String>,
     pub max_lsn_wal_lag: Option<NonZeroU64>,
+    pub trace_read_requests: Option<bool>,
 }
 
 impl TenantConfigRequest {
@@ -109,6 +132,7 @@ impl TenantConfigRequest {
             walreceiver_connect_timeout: None,
             lagging_wal_timeout: None,
             max_lsn_wal_lag: None,
+            trace_read_requests: None,
         }
     }
 }
@@ -160,6 +184,8 @@ pub struct TimelineInfo {
     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,
@@ -201,3 +227,262 @@ pub struct FailpointConfig {
 pub struct TimelineGcRequest {
     pub gc_horizon: Option<u64>,
 }
+
+// Wrapped in libpq CopyData
+#[derive(PartialEq, Eq)]
+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, PartialEq, Eq)]
+pub struct PagestreamExistsRequest {
+    pub latest: bool,
+    pub lsn: Lsn,
+    pub rel: RelTag,
+}
+
+#[derive(Debug, PartialEq, Eq)]
+pub struct PagestreamNblocksRequest {
+    pub latest: bool,
+    pub lsn: Lsn,
+    pub rel: RelTag,
+}
+
+#[derive(Debug, PartialEq, Eq)]
+pub struct PagestreamGetPageRequest {
+    pub latest: bool,
+    pub lsn: Lsn,
+    pub rel: RelTag,
+    pub blkno: u32,
+}
+
+#[derive(Debug, PartialEq, Eq)]
+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 serialize(&self) -> Bytes {
+        let mut bytes = BytesMut::new();
+
+        match self {
+            Self::Exists(req) => {
+                bytes.put_u8(0);
+                bytes.put_u8(if req.latest { 1 } else { 0 });
+                bytes.put_u64(req.lsn.0);
+                bytes.put_u32(req.rel.spcnode);
+                bytes.put_u32(req.rel.dbnode);
+                bytes.put_u32(req.rel.relnode);
+                bytes.put_u8(req.rel.forknum);
+            }
+
+            Self::Nblocks(req) => {
+                bytes.put_u8(1);
+                bytes.put_u8(if req.latest { 1 } else { 0 });
+                bytes.put_u64(req.lsn.0);
+                bytes.put_u32(req.rel.spcnode);
+                bytes.put_u32(req.rel.dbnode);
+                bytes.put_u32(req.rel.relnode);
+                bytes.put_u8(req.rel.forknum);
+            }
+
+            Self::GetPage(req) => {
+                bytes.put_u8(2);
+                bytes.put_u8(if req.latest { 1 } else { 0 });
+                bytes.put_u64(req.lsn.0);
+                bytes.put_u32(req.rel.spcnode);
+                bytes.put_u32(req.rel.dbnode);
+                bytes.put_u32(req.rel.relnode);
+                bytes.put_u8(req.rel.forknum);
+                bytes.put_u32(req.blkno);
+            }
+
+            Self::DbSize(req) => {
+                bytes.put_u8(3);
+                bytes.put_u8(if req.latest { 1 } else { 0 });
+                bytes.put_u64(req.lsn.0);
+                bytes.put_u32(req.dbnode);
+            }
+        }
+
+        bytes.into()
+    }
+
+    pub fn parse<R: std::io::Read>(body: &mut R) -> 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.read_u8()?;
+        match msg_tag {
+            0 => Ok(PagestreamFeMessage::Exists(PagestreamExistsRequest {
+                latest: body.read_u8()? != 0,
+                lsn: Lsn::from(body.read_u64::<BigEndian>()?),
+                rel: RelTag {
+                    spcnode: body.read_u32::<BigEndian>()?,
+                    dbnode: body.read_u32::<BigEndian>()?,
+                    relnode: body.read_u32::<BigEndian>()?,
+                    forknum: body.read_u8()?,
+                },
+            })),
+            1 => Ok(PagestreamFeMessage::Nblocks(PagestreamNblocksRequest {
+                latest: body.read_u8()? != 0,
+                lsn: Lsn::from(body.read_u64::<BigEndian>()?),
+                rel: RelTag {
+                    spcnode: body.read_u32::<BigEndian>()?,
+                    dbnode: body.read_u32::<BigEndian>()?,
+                    relnode: body.read_u32::<BigEndian>()?,
+                    forknum: body.read_u8()?,
+                },
+            })),
+            2 => Ok(PagestreamFeMessage::GetPage(PagestreamGetPageRequest {
+                latest: body.read_u8()? != 0,
+                lsn: Lsn::from(body.read_u64::<BigEndian>()?),
+                rel: RelTag {
+                    spcnode: body.read_u32::<BigEndian>()?,
+                    dbnode: body.read_u32::<BigEndian>()?,
+                    relnode: body.read_u32::<BigEndian>()?,
+                    forknum: body.read_u8()?,
+                },
+                blkno: body.read_u32::<BigEndian>()?,
+            })),
+            3 => Ok(PagestreamFeMessage::DbSize(PagestreamDbSizeRequest {
+                latest: body.read_u8()? != 0,
+                lsn: Lsn::from(body.read_u64::<BigEndian>()?),
+                dbnode: body.read_u32::<BigEndian>()?,
+            })),
+            _ => bail!("unknown smgr message tag: {:?}", msg_tag),
+        }
+    }
+}
+
+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()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use bytes::Buf;
+
+    use super::*;
+
+    #[test]
+    fn test_pagestream() {
+        // Test serialization/deserialization of PagestreamFeMessage
+        let messages = vec![
+            PagestreamFeMessage::Exists(PagestreamExistsRequest {
+                latest: true,
+                lsn: Lsn(4),
+                rel: RelTag {
+                    forknum: 1,
+                    spcnode: 2,
+                    dbnode: 3,
+                    relnode: 4,
+                },
+            }),
+            PagestreamFeMessage::Nblocks(PagestreamNblocksRequest {
+                latest: false,
+                lsn: Lsn(4),
+                rel: RelTag {
+                    forknum: 1,
+                    spcnode: 2,
+                    dbnode: 3,
+                    relnode: 4,
+                },
+            }),
+            PagestreamFeMessage::GetPage(PagestreamGetPageRequest {
+                latest: true,
+                lsn: Lsn(4),
+                rel: RelTag {
+                    forknum: 1,
+                    spcnode: 2,
+                    dbnode: 3,
+                    relnode: 4,
+                },
+                blkno: 7,
+            }),
+            PagestreamFeMessage::DbSize(PagestreamDbSizeRequest {
+                latest: true,
+                lsn: Lsn(4),
+                dbnode: 7,
+            }),
+        ];
+        for msg in messages {
+            let bytes = msg.serialize();
+            let reconstructed = PagestreamFeMessage::parse(&mut bytes.reader()).unwrap();
+            assert!(msg == reconstructed);
+        }
+    }
+}

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 = "0.6.2"
+memoffset = "0.7"
 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 = "0.60.1"
+bindgen = "0.61"

libs/postgres_ffi/wal_craft/Cargo.toml

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

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

@@ -1,68 +1,19 @@
 use anyhow::*;
-use clap::{App, Arg, ArgMatches};
-use std::str::FromStr;
+use clap::{value_parser, Arg, ArgMatches, Command};
+use std::{path::PathBuf, 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 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 arg_matches = cli().get_matches();
 
     let wal_craft = |arg_matches: &ArgMatches, client| {
-        let (intermediate_lsns, end_of_wal_lsn) = match arg_matches.value_of("type").unwrap() {
+        let (intermediate_lsns, end_of_wal_lsn) = match arg_matches
+            .get_one::<String>("type")
+            .map(|s| s.as_str())
+            .context("'type' is required")?
+        {
             Simple::NAME => Simple::craft(client)?,
             LastWalRecordXlogSwitch::NAME => LastWalRecordXlogSwitch::craft(client)?,
             LastWalRecordXlogSwitchEndsOnPageBoundary::NAME => {
@@ -72,12 +23,12 @@ 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(())
     };
 
@@ -85,20 +36,24 @@ 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
-                    .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(),
+                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(),
             };
             cfg.initdb()?;
             let srv = cfg.start_server()?;
@@ -108,9 +63,77 @@ fn main() -> Result<()> {
         }
         Some(("in-existing", arg_matches)) => wal_craft(
             arg_matches,
-            &mut postgres::Config::from_str(arg_matches.value_of("connection").unwrap())?
-                .connect(postgres::NoTls)?,
+            &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)?,
         ),
         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) -> PathBuf {
+    pub fn pg_distrib_dir(&self) -> anyhow::Result<PathBuf> {
         let path = self.pg_distrib_dir.clone();
 
         match 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),
+            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),
         }
     }
 
-    fn pg_bin_dir(&self) -> PathBuf {
-        self.pg_distrib_dir().join("bin")
+    fn pg_bin_dir(&self) -> anyhow::Result<PathBuf> {
+        Ok(self.pg_distrib_dir()?.join("bin"))
     }
 
-    fn pg_lib_dir(&self) -> PathBuf {
-        self.pg_distrib_dir().join("lib")
+    fn pg_lib_dir(&self) -> anyhow::Result<PathBuf> {
+        Ok(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/pq_proto/Cargo.toml

@@ -0,0 +1,16 @@
+[package]
+name = "pq_proto"
+version = "0.1.0"
+edition = "2021"
+
+[dependencies]
+anyhow = "1.0"
+bytes = "1.0.1"
+pin-project-lite = "0.2.7"
+postgres-protocol = { git = "https://github.com/neondatabase/rust-postgres.git", rev="d052ee8b86fff9897c77b0fe89ea9daba0e1fa38" }
+rand = "0.8.3"
+serde = { version = "1.0", features = ["derive"] }
+tokio = { version = "1.17", features = ["macros"] }
+tracing = "0.1"
+
+workspace_hack = { version = "0.1", path = "../../workspace_hack" }

libs/pq_proto/src/lib.rs

@@ -2,7 +2,9 @@
 //! <https://www.postgresql.org/docs/devel/protocol-message-formats.html>
 //! on message formats.
 
-use crate::sync::{AsyncishRead, SyncFuture};
+// Tools for calling certain async methods in sync contexts.
+pub mod sync;
+
 use anyhow::{bail, ensure, Context, Result};
 use bytes::{Buf, BufMut, Bytes, BytesMut};
 use postgres_protocol::PG_EPOCH;
@@ -16,6 +18,7 @@ use std::{
     str,
     time::{Duration, SystemTime},
 };
+use sync::{AsyncishRead, SyncFuture};
 use tokio::io::AsyncReadExt;
 use tracing::{trace, warn};
 
@@ -198,7 +201,7 @@ impl FeMessage {
     ///
     /// ```
     /// # use std::io;
-    /// # use utils::pq_proto::FeMessage;
+    /// # use pq_proto::FeMessage;
     /// #
     /// # fn process_message(msg: FeMessage) -> anyhow::Result<()> {
     /// #     Ok(())
@@ -302,6 +305,7 @@ impl FeStartupPacket {
                 Err(e) => return Err(e.into()),
             };
 
+            #[allow(clippy::manual_range_contains)]
             if len < 4 || len > MAX_STARTUP_PACKET_LENGTH {
                 bail!("invalid message length");
             }

libs/pq_proto/src/sync.rs

@@ -29,7 +29,7 @@ impl<S, T: Future> SyncFuture<S, T> {
     /// Example:
     ///
     /// ```
-    /// # use utils::sync::SyncFuture;
+    /// # use pq_proto::sync::SyncFuture;
     /// # use std::future::Future;
     /// # use tokio::io::AsyncReadExt;
     /// #

libs/remote_storage/src/local_fs.rs

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

libs/tenant_size_model/.gitignore

@@ -0,0 +1,3 @@
+*.dot
+*.png
+*.svg

libs/tenant_size_model/Cargo.toml

@@ -0,0 +1,8 @@
+[package]
+name = "tenant_size_model"
+version = "0.1.0"
+edition = "2021"
+publish = false
+
+[dependencies]
+workspace_hack = { version = "0.1", path = "../../workspace_hack" }

libs/tenant_size_model/Makefile

@@ -0,0 +1,13 @@
+all: 1.svg 2.svg 3.svg 4.svg 1.png 2.png 3.png 4.png
+
+../../target/debug/tenant_size_model: Cargo.toml src/main.rs src/lib.rs
+	cargo build --bin tenant_size_model
+
+%.svg: %.dot
+	dot -Tsvg $< > $@
+
+%.png: %.dot
+	dot -Tpng $< > $@
+
+%.dot: ../../target/debug/tenant_size_model
+	../../target/debug/tenant_size_model $* > $@

libs/tenant_size_model/README.md

@@ -0,0 +1,7 @@
+# Logical size + WAL pricing
+
+This is a simulator to calculate the tenant size in different scenarios,
+using the "Logical size + WAL" method. Makefile produces diagrams used in a
+private presentation:
+
+https://docs.google.com/presentation/d/1OapE4k11xmcwMh7I7YvNWGC63yCRLh6udO9bXZ-fZmo/edit?usp=sharing

libs/tenant_size_model/src/lib.rs

@@ -0,0 +1,349 @@
+use std::borrow::Cow;
+use std::collections::HashMap;
+
+/// Pricing model or history size builder.
+///
+/// Maintains knowledge of the branches and their modifications. Generic over the branch name key
+/// type.
+pub struct Storage<K: 'static> {
+    segments: Vec<Segment>,
+
+    /// Mapping from the branch name to the index of a segment describing it's latest state.
+    branches: HashMap<K, usize>,
+}
+
+/// Snapshot of a branch.
+#[derive(Clone, Debug, Eq, PartialEq)]
+pub struct Segment {
+    /// Previous segment index into ['Storage::segments`], if any.
+    parent: Option<usize>,
+
+    /// Description of how did we get to this state.
+    ///
+    /// Mainly used in the original scenarios 1..=4 with insert, delete and update. Not used when
+    /// modifying a branch directly.
+    pub op: Cow<'static, str>,
+
+    /// LSN before this state
+    start_lsn: u64,
+
+    /// LSN at this state
+    pub end_lsn: u64,
+
+    /// Logical size before this state
+    start_size: u64,
+
+    /// Logical size at this state
+    pub end_size: u64,
+
+    /// Indices to [`Storage::segments`]
+    ///
+    /// FIXME: this could be an Option<usize>
+    children_after: Vec<usize>,
+
+    /// Determined by `retention_period` given to [`Storage::calculate`]
+    pub needed: bool,
+}
+
+//
+//
+//
+//
+//                 *-g--*---D--->
+//                /
+//               /
+//              /                 *---b----*-B--->
+//             /                 /
+//            /                 /
+//      -----*--e---*-----f----* C
+//           E                  \
+//                               \
+//                                *--a---*---A-->
+//
+// If A and B need to be retained, is it cheaper to store
+// snapshot at C+a+b, or snapshots at A and B ?
+//
+// If D also needs to be retained, which is cheaper:
+//
+// 1. E+g+e+f+a+b
+// 2. D+C+a+b
+// 3. D+A+B
+
+/// [`Segment`] which has had it's size calculated.
+pub struct SegmentSize {
+    pub seg_id: usize,
+
+    pub method: SegmentMethod,
+
+    this_size: u64,
+
+    pub children: Vec<SegmentSize>,
+}
+
+impl SegmentSize {
+    fn total(&self) -> u64 {
+        self.this_size + self.children.iter().fold(0, |acc, x| acc + x.total())
+    }
+
+    pub fn total_children(&self) -> u64 {
+        if self.method == SnapshotAfter {
+            self.this_size + self.children.iter().fold(0, |acc, x| acc + x.total())
+        } else {
+            self.children.iter().fold(0, |acc, x| acc + x.total())
+        }
+    }
+}
+
+/// Different methods to retain history from a particular state
+#[derive(Clone, Copy, Debug, Eq, PartialEq)]
+pub enum SegmentMethod {
+    SnapshotAfter,
+    Wal,
+    WalNeeded,
+    Skipped,
+}
+
+use SegmentMethod::*;
+
+impl<K: std::hash::Hash + Eq + 'static> Storage<K> {
+    /// Creates a new storage with the given default branch name.
+    pub fn new(initial_branch: K) -> Storage<K> {
+        let init_segment = Segment {
+            op: "".into(),
+            needed: false,
+            parent: None,
+            start_lsn: 0,
+            end_lsn: 0,
+            start_size: 0,
+            end_size: 0,
+            children_after: Vec::new(),
+        };
+
+        Storage {
+            segments: vec![init_segment],
+            branches: HashMap::from([(initial_branch, 0)]),
+        }
+    }
+
+    /// Advances the branch with the named operation, by the relative LSN and logical size bytes.
+    pub fn modify_branch<Q: ?Sized>(
+        &mut self,
+        branch: &Q,
+        op: Cow<'static, str>,
+        lsn_bytes: u64,
+        size_bytes: i64,
+    ) where
+        K: std::borrow::Borrow<Q>,
+        Q: std::hash::Hash + Eq,
+    {
+        let lastseg_id = *self.branches.get(branch).unwrap();
+        let newseg_id = self.segments.len();
+        let lastseg = &mut self.segments[lastseg_id];
+
+        let newseg = Segment {
+            op,
+            parent: Some(lastseg_id),
+            start_lsn: lastseg.end_lsn,
+            end_lsn: lastseg.end_lsn + lsn_bytes,
+            start_size: lastseg.end_size,
+            end_size: (lastseg.end_size as i64 + size_bytes) as u64,
+            children_after: Vec::new(),
+            needed: false,
+        };
+        lastseg.children_after.push(newseg_id);
+
+        self.segments.push(newseg);
+        *self.branches.get_mut(branch).expect("read already") = newseg_id;
+    }
+
+    pub fn insert<Q: ?Sized>(&mut self, branch: &Q, bytes: u64)
+    where
+        K: std::borrow::Borrow<Q>,
+        Q: std::hash::Hash + Eq,
+    {
+        self.modify_branch(branch, "insert".into(), bytes, bytes as i64);
+    }
+
+    pub fn update<Q: ?Sized>(&mut self, branch: &Q, bytes: u64)
+    where
+        K: std::borrow::Borrow<Q>,
+        Q: std::hash::Hash + Eq,
+    {
+        self.modify_branch(branch, "update".into(), bytes, 0i64);
+    }
+
+    pub fn delete<Q: ?Sized>(&mut self, branch: &Q, bytes: u64)
+    where
+        K: std::borrow::Borrow<Q>,
+        Q: std::hash::Hash + Eq,
+    {
+        self.modify_branch(branch, "delete".into(), bytes, -(bytes as i64));
+    }
+
+    /// Panics if the parent branch cannot be found.
+    pub fn branch<Q: ?Sized>(&mut self, parent: &Q, name: K)
+    where
+        K: std::borrow::Borrow<Q>,
+        Q: std::hash::Hash + Eq,
+    {
+        // Find the right segment
+        let branchseg_id = *self
+            .branches
+            .get(parent)
+            .expect("should had found the parent by key");
+        let _branchseg = &mut self.segments[branchseg_id];
+
+        // Create branch name for it
+        self.branches.insert(name, branchseg_id);
+    }
+
+    pub fn calculate(&mut self, retention_period: u64) -> SegmentSize {
+        // Phase 1: Mark all the segments that need to be retained
+        for (_branch, &last_seg_id) in self.branches.iter() {
+            let last_seg = &self.segments[last_seg_id];
+            let cutoff_lsn = last_seg.start_lsn.saturating_sub(retention_period);
+            let mut seg_id = last_seg_id;
+            loop {
+                let seg = &mut self.segments[seg_id];
+                if seg.end_lsn < cutoff_lsn {
+                    break;
+                }
+                seg.needed = true;
+                if let Some(prev_seg_id) = seg.parent {
+                    seg_id = prev_seg_id;
+                } else {
+                    break;
+                }
+            }
+        }
+
+        // Phase 2: For each oldest segment in a chain that needs to be retained,
+        // calculate if we should store snapshot or WAL
+        self.size_from_snapshot_later(0)
+    }
+
+    fn size_from_wal(&self, seg_id: usize) -> SegmentSize {
+        let seg = &self.segments[seg_id];
+
+        let this_size = seg.end_lsn - seg.start_lsn;
+
+        let mut children = Vec::new();
+
+        // try both ways
+        for &child_id in seg.children_after.iter() {
+            // try each child both ways
+            let child = &self.segments[child_id];
+            let p1 = self.size_from_wal(child_id);
+
+            let p = if !child.needed {
+                let p2 = self.size_from_snapshot_later(child_id);
+                if p1.total() < p2.total() {
+                    p1
+                } else {
+                    p2
+                }
+            } else {
+                p1
+            };
+            children.push(p);
+        }
+        SegmentSize {
+            seg_id,
+            method: if seg.needed { WalNeeded } else { Wal },
+            this_size,
+            children,
+        }
+    }
+
+    fn size_from_snapshot_later(&self, seg_id: usize) -> SegmentSize {
+        // If this is needed, then it's time to do the snapshot and continue
+        // with wal method.
+        let seg = &self.segments[seg_id];
+        //eprintln!("snap: seg{}: {} needed: {}", seg_id, seg.children_after.len(), seg.needed);
+        if seg.needed {
+            let mut children = Vec::new();
+
+            for &child_id in seg.children_after.iter() {
+                // try each child both ways
+                let child = &self.segments[child_id];
+                let p1 = self.size_from_wal(child_id);
+
+                let p = if !child.needed {
+                    let p2 = self.size_from_snapshot_later(child_id);
+                    if p1.total() < p2.total() {
+                        p1
+                    } else {
+                        p2
+                    }
+                } else {
+                    p1
+                };
+                children.push(p);
+            }
+            SegmentSize {
+                seg_id,
+                method: WalNeeded,
+                this_size: seg.start_size,
+                children,
+            }
+        } else {
+            // If any of the direct children are "needed", need to be able to reconstruct here
+            let mut children_needed = false;
+            for &child in seg.children_after.iter() {
+                let seg = &self.segments[child];
+                if seg.needed {
+                    children_needed = true;
+                    break;
+                }
+            }
+
+            let method1 = if !children_needed {
+                let mut children = Vec::new();
+                for child in seg.children_after.iter() {
+                    children.push(self.size_from_snapshot_later(*child));
+                }
+                Some(SegmentSize {
+                    seg_id,
+                    method: Skipped,
+                    this_size: 0,
+                    children,
+                })
+            } else {
+                None
+            };
+
+            // If this a junction, consider snapshotting here
+            let method2 = if children_needed || seg.children_after.len() >= 2 {
+                let mut children = Vec::new();
+                for child in seg.children_after.iter() {
+                    children.push(self.size_from_wal(*child));
+                }
+                Some(SegmentSize {
+                    seg_id,
+                    method: SnapshotAfter,
+                    this_size: seg.end_size,
+                    children,
+                })
+            } else {
+                None
+            };
+
+            match (method1, method2) {
+                (None, None) => panic!(),
+                (Some(method), None) => method,
+                (None, Some(method)) => method,
+                (Some(method1), Some(method2)) => {
+                    if method1.total() < method2.total() {
+                        method1
+                    } else {
+                        method2
+                    }
+                }
+            }
+        }
+    }
+
+    pub fn into_segments(self) -> Vec<Segment> {
+        self.segments
+    }
+}

libs/tenant_size_model/src/main.rs

@@ -0,0 +1,268 @@
+//! Tenant size model testing ground.
+//!
+//! Has a number of scenarios and a `main` for invoking these by number, calculating the history
+//! size, outputs graphviz graph. Makefile in directory shows how to use graphviz to turn scenarios
+//! into pngs.
+
+use tenant_size_model::{Segment, SegmentSize, Storage};
+
+// Main branch only. Some updates on it.
+fn scenario_1() -> (Vec<Segment>, SegmentSize) {
+    // Create main branch
+    let mut storage = Storage::new("main");
+
+    // Bulk load 5 GB of data to it
+    storage.insert("main", 5_000);
+
+    // Stream of updates
+    for _ in 0..5 {
+        storage.update("main", 1_000);
+    }
+
+    let size = storage.calculate(1000);
+
+    (storage.into_segments(), size)
+}
+
+// Main branch only. Some updates on it.
+fn scenario_2() -> (Vec<Segment>, SegmentSize) {
+    // Create main branch
+    let mut storage = Storage::new("main");
+
+    // Bulk load 5 GB of data to it
+    storage.insert("main", 5_000);
+
+    // Stream of updates
+    for _ in 0..5 {
+        storage.update("main", 1_000);
+    }
+
+    // Branch
+    storage.branch("main", "child");
+    storage.update("child", 1_000);
+
+    // More updates on parent
+    storage.update("main", 1_000);
+
+    let size = storage.calculate(1000);
+
+    (storage.into_segments(), size)
+}
+
+// Like 2, but more updates on main
+fn scenario_3() -> (Vec<Segment>, SegmentSize) {
+    // Create main branch
+    let mut storage = Storage::new("main");
+
+    // Bulk load 5 GB of data to it
+    storage.insert("main", 5_000);
+
+    // Stream of updates
+    for _ in 0..5 {
+        storage.update("main", 1_000);
+    }
+
+    // Branch
+    storage.branch("main", "child");
+    storage.update("child", 1_000);
+
+    // More updates on parent
+    for _ in 0..5 {
+        storage.update("main", 1_000);
+    }
+
+    let size = storage.calculate(1000);
+
+    (storage.into_segments(), size)
+}
+
+// Diverged branches
+fn scenario_4() -> (Vec<Segment>, SegmentSize) {
+    // Create main branch
+    let mut storage = Storage::new("main");
+
+    // Bulk load 5 GB of data to it
+    storage.insert("main", 5_000);
+
+    // Stream of updates
+    for _ in 0..5 {
+        storage.update("main", 1_000);
+    }
+
+    // Branch
+    storage.branch("main", "child");
+    storage.update("child", 1_000);
+
+    // More updates on parent
+    for _ in 0..8 {
+        storage.update("main", 1_000);
+    }
+
+    let size = storage.calculate(1000);
+
+    (storage.into_segments(), size)
+}
+
+fn scenario_5() -> (Vec<Segment>, SegmentSize) {
+    let mut storage = Storage::new("a");
+    storage.insert("a", 5000);
+    storage.branch("a", "b");
+    storage.update("b", 4000);
+    storage.update("a", 2000);
+    storage.branch("a", "c");
+    storage.insert("c", 4000);
+    storage.insert("a", 2000);
+
+    let size = storage.calculate(5000);
+
+    (storage.into_segments(), size)
+}
+
+fn scenario_6() -> (Vec<Segment>, SegmentSize) {
+    use std::borrow::Cow;
+
+    const NO_OP: Cow<'static, str> = Cow::Borrowed("");
+
+    let branches = [
+        Some(0x7ff1edab8182025f15ae33482edb590a_u128),
+        Some(0xb1719e044db05401a05a2ed588a3ad3f),
+        Some(0xb68d6691c895ad0a70809470020929ef),
+    ];
+
+    // compared to other scenarios, this one uses bytes instead of kB
+
+    let mut storage = Storage::new(None);
+
+    storage.branch(&None, branches[0]); // at 0
+    storage.modify_branch(&branches[0], NO_OP, 108951064, 43696128); // at 108951064
+    storage.branch(&branches[0], branches[1]); // at 108951064
+    storage.modify_branch(&branches[1], NO_OP, 15560408, -1851392); // at 124511472
+    storage.modify_branch(&branches[0], NO_OP, 174464360, -1531904); // at 283415424
+    storage.branch(&branches[0], branches[2]); // at 283415424
+    storage.modify_branch(&branches[2], NO_OP, 15906192, 8192); // at 299321616
+    storage.modify_branch(&branches[0], NO_OP, 18909976, 32768); // at 302325400
+
+    let size = storage.calculate(100_000);
+
+    (storage.into_segments(), size)
+}
+
+fn main() {
+    let args: Vec<String> = std::env::args().collect();
+
+    let scenario = if args.len() < 2 { "1" } else { &args[1] };
+
+    let (segments, size) = match scenario {
+        "1" => scenario_1(),
+        "2" => scenario_2(),
+        "3" => scenario_3(),
+        "4" => scenario_4(),
+        "5" => scenario_5(),
+        "6" => scenario_6(),
+        other => {
+            eprintln!("invalid scenario {}", other);
+            std::process::exit(1);
+        }
+    };
+
+    graphviz_tree(&segments, &size);
+}
+
+fn graphviz_recurse(segments: &[Segment], node: &SegmentSize) {
+    use tenant_size_model::SegmentMethod::*;
+
+    let seg_id = node.seg_id;
+    let seg = segments.get(seg_id).unwrap();
+    let lsn = seg.end_lsn;
+    let size = seg.end_size;
+    let method = node.method;
+
+    println!("  {{");
+    println!("    node [width=0.1 height=0.1 shape=oval]");
+
+    let tenant_size = node.total_children();
+
+    let penwidth = if seg.needed { 6 } else { 3 };
+    let x = match method {
+        SnapshotAfter =>
+            format!("label=\"lsn: {lsn}\\nsize: {size}\\ntenant_size: {tenant_size}\" style=filled penwidth={penwidth}"),
+        Wal =>
+            format!("label=\"lsn: {lsn}\\nsize: {size}\\ntenant_size: {tenant_size}\" color=\"black\" penwidth={penwidth}"),
+        WalNeeded =>
+            format!("label=\"lsn: {lsn}\\nsize: {size}\\ntenant_size: {tenant_size}\" color=\"black\" penwidth={penwidth}"),
+        Skipped =>
+            format!("label=\"lsn: {lsn}\\nsize: {size}\\ntenant_size: {tenant_size}\" color=\"gray\" penwidth={penwidth}"),
+    };
+
+    println!("    \"seg{seg_id}\" [{x}]");
+    println!("  }}");
+
+    // Recurse. Much of the data is actually on the edge
+    for child in node.children.iter() {
+        let child_id = child.seg_id;
+        graphviz_recurse(segments, child);
+
+        let edge_color = match child.method {
+            SnapshotAfter => "gray",
+            Wal => "black",
+            WalNeeded => "black",
+            Skipped => "gray",
+        };
+
+        println!("  {{");
+        println!("    edge [] ");
+        print!("    \"seg{seg_id}\" -> \"seg{child_id}\" [");
+        print!("color={edge_color}");
+        if child.method == WalNeeded {
+            print!(" penwidth=6");
+        }
+        if child.method == Wal {
+            print!(" penwidth=3");
+        }
+
+        let next = segments.get(child_id).unwrap();
+
+        if next.op.is_empty() {
+            print!(
+                " label=\"{} / {}\"",
+                next.end_lsn - seg.end_lsn,
+                (next.end_size as i128 - seg.end_size as i128)
+            );
+        } else {
+            print!(" label=\"{}: {}\"", next.op, next.end_lsn - seg.end_lsn);
+        }
+        println!("]");
+        println!("  }}");
+    }
+}
+
+fn graphviz_tree(segments: &[Segment], tree: &SegmentSize) {
+    println!("digraph G {{");
+    println!("  fontname=\"Helvetica,Arial,sans-serif\"");
+    println!("  node [fontname=\"Helvetica,Arial,sans-serif\"]");
+    println!("  edge [fontname=\"Helvetica,Arial,sans-serif\"]");
+    println!("  graph [center=1 rankdir=LR]");
+    println!("  edge [dir=none]");
+
+    graphviz_recurse(segments, tree);
+
+    println!("}}");
+}
+
+#[test]
+fn scenarios_return_same_size() {
+    type ScenarioFn = fn() -> (Vec<Segment>, SegmentSize);
+    let truths: &[(u32, ScenarioFn, _)] = &[
+        (line!(), scenario_1, 8000),
+        (line!(), scenario_2, 9000),
+        (line!(), scenario_3, 13000),
+        (line!(), scenario_4, 16000),
+        (line!(), scenario_5, 17000),
+        (line!(), scenario_6, 333_792_000),
+    ];
+
+    for (line, scenario, expected) in truths {
+        let (_, size) = scenario();
+        assert_eq!(*expected, size.total_children(), "scenario on line {line}");
+    }
+}

libs/utils/Cargo.toml

@@ -9,9 +9,6 @@ anyhow = "1.0"
 bincode = "1.3"
 bytes = "1.0.1"
 hyper = { version = "0.14.7", features = ["full"] }
-pin-project-lite = "0.2.7"
-postgres = { git = "https://github.com/neondatabase/rust-postgres.git", rev="d052ee8b86fff9897c77b0fe89ea9daba0e1fa38" }
-postgres-protocol = { git = "https://github.com/neondatabase/rust-postgres.git", rev="d052ee8b86fff9897c77b0fe89ea9daba0e1fa38" }
 routerify = "3"
 serde = { version = "1.0", features = ["derive"] }
 serde_json = "1"
@@ -19,7 +16,7 @@ thiserror = "1.0"
 tokio = { version = "1.17", features = ["macros"]}
 tokio-rustls = "0.23"
 tracing = "0.1"
-tracing-subscriber = { version = "0.3", features = ["env-filter"] }
+tracing-subscriber = { version = "0.3", features = ["env-filter", "json"] }
 nix = "0.25"
 signal-hook = "0.3.10"
 rand = "0.8.3"
@@ -30,9 +27,11 @@ rustls-split = "0.3.0"
 git-version = "0.3.5"
 serde_with = "2.0"
 once_cell = "1.13.0"
-
+strum = "0.24"
+strum_macros = "0.24"
 
 metrics = { path = "../metrics" }
+pq_proto = { path = "../pq_proto" }
 workspace_hack = { version = "0.1", path = "../../workspace_hack" }
 
 [dev-dependencies]

libs/utils/src/connstring.rs

@@ -1,52 +0,0 @@
-use postgres::Config;
-
-pub fn connection_host_port(config: &Config) -> (String, u16) {
-    assert_eq!(
-        config.get_hosts().len(),
-        1,
-        "only one pair of host and port is supported in connection string"
-    );
-    assert_eq!(
-        config.get_ports().len(),
-        1,
-        "only one pair of host and port is supported in connection string"
-    );
-    let host = match &config.get_hosts()[0] {
-        postgres::config::Host::Tcp(host) => host.as_ref(),
-        postgres::config::Host::Unix(host) => host.to_str().unwrap(),
-    };
-    (host.to_owned(), config.get_ports()[0])
-}
-
-pub fn connection_address(config: &Config) -> String {
-    let (host, port) = connection_host_port(config);
-    format!("{}:{}", host, port)
-}
-
-#[cfg(test)]
-mod tests {
-    use super::*;
-
-    #[test]
-    fn test_connection_host_port() {
-        let config: Config = "postgresql://no_user@localhost:64000/no_db"
-            .parse()
-            .unwrap();
-        assert_eq!(
-            connection_host_port(&config),
-            ("localhost".to_owned(), 64000)
-        );
-    }
-
-    #[test]
-    #[should_panic(expected = "only one pair of host and port is supported in connection string")]
-    fn test_connection_host_port_multiple_ports() {
-        let config: Config = "postgresql://no_user@localhost:64000,localhost:64001/no_db"
-            .parse()
-            .unwrap();
-        assert_eq!(
-            connection_host_port(&config),
-            ("localhost".to_owned(), 64000)
-        );
-    }
-}

libs/utils/src/crashsafe.rs

@@ -12,16 +12,8 @@ pub fn create_dir(path: impl AsRef<Path>) -> io::Result<()> {
     let path = path.as_ref();
 
     fs::create_dir(path)?;
-    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",
-        ))
-    }
+    fsync_file_and_parent(path)?;
+    Ok(())
 }
 
 /// Similar to [`std::fs::create_dir_all`], except we fsync all
@@ -65,12 +57,12 @@ 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() {
-        File::open(path)?.sync_all()?;
+        fsync(path)?;
     }
 
     // If we created any new directories, fsync the parent.
     if !dirs_to_create.is_empty() {
-        File::open(path)?.sync_all()?;
+        fsync(path)?;
     }
 
     Ok(())
@@ -92,6 +84,33 @@ 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

@@ -8,7 +8,7 @@ use thiserror::Error;
 #[derive(Error, Debug)]
 pub enum IdError {
     #[error("invalid id length {0}")]
-    VecParseError(usize),
+    SliceParseError(usize),
 }
 
 /// Neon ID is a 128-bit random ID.
@@ -29,13 +29,13 @@ impl Id {
         Id::from(arr)
     }
 
-    pub fn from_vec(src: &Vec<u8>) -> Result<Id, IdError> {
+    pub fn from_slice(src: &[u8]) -> Result<Id, IdError> {
         if src.len() != 16 {
-            return Err(IdError::VecParseError(src.len()));
+            return Err(IdError::SliceParseError(src.len()));
         }
-        let mut zid_slice = [0u8; 16];
-        zid_slice.copy_from_slice(&src);
-        Ok(zid_slice.into())
+        let mut id_array = [0u8; 16];
+        id_array.copy_from_slice(src);
+        Ok(id_array.into())
     }
 
     pub fn as_arr(&self) -> [u8; 16] {
@@ -116,8 +116,8 @@ macro_rules! id_newtype {
                 $t(Id::get_from_buf(buf))
             }
 
-            pub fn from_vec(src: &Vec<u8>) -> Result<$t, IdError> {
-                Ok($t(Id::from_vec(src)?))
+            pub fn from_slice(src: &[u8]) -> Result<$t, IdError> {
+                Ok($t(Id::from_slice(src)?))
             }
 
             pub fn as_arr(&self) -> [u8; 16] {
@@ -224,6 +224,17 @@ pub struct TenantId(Id);
 
 id_newtype!(TenantId);
 
+/// Neon Connection Id identifies long-lived connections (for example a pagestream
+/// connection with the page_service). Is used for better logging and tracing
+///
+/// NOTE: It (de)serializes as an array of hex bytes, so the string representation would look
+/// like `[173,80,132,115,129,226,72,254,170,201,135,108,199,26,228,24]`.
+/// See [`Id`] for alternative ways to serialize it.
+#[derive(Clone, Copy, PartialEq, Eq, Hash, Serialize, Deserialize, PartialOrd, Ord)]
+pub struct ConnectionId(Id);
+
+id_newtype!(ConnectionId);
+
 // A pair uniquely identifying Neon instance.
 #[derive(Debug, Clone, Copy, PartialOrd, Ord, PartialEq, Eq, Hash, Serialize, Deserialize)]
 pub struct TenantTimelineId {

libs/utils/src/lib.rs

@@ -1,8 +1,6 @@
 //! `utils` is intended to be a place to put code that is shared
 //! between other crates in this repository.
 
-#![allow(clippy::manual_range_contains)]
-
 /// `Lsn` type implements common tasks on Log Sequence Numbers
 pub mod lsn;
 /// SeqWait allows waiting for a future sequence number to arrive
@@ -17,13 +15,9 @@ pub mod vec_map;
 pub mod bin_ser;
 pub mod postgres_backend;
 pub mod postgres_backend_async;
-pub mod pq_proto;
 
-// dealing with connstring parsing and handy access to it's parts
-pub mod connstring;
-
-// helper functions for creating and fsyncing directories/trees
-pub mod crashsafe_dir;
+// helper functions for creating and fsyncing
+pub mod crashsafe;
 
 // common authentication routines
 pub mod auth;
@@ -39,13 +33,12 @@ pub mod sock_split;
 // common log initialisation routine
 pub mod logging;
 
+pub mod lock_file;
+
 // Misc
 pub mod accum;
 pub mod shutdown;
 
-// Tools for calling certain async methods in sync contexts
-pub mod sync;
-
 // Utility for binding TcpListeners with proper socket options.
 pub mod tcp_listener;
 

libs/utils/src/lock_file.rs

@@ -0,0 +1,81 @@
+//! A module to create and read lock files. A lock file ensures that only one
+//! process is running at a time, in a particular directory.
+//!
+//! File locking is done using [`fcntl::flock`], which means that holding the
+//! lock on file only prevents acquiring another lock on it; all other
+//! operations are still possible on files. Other process can still open, read,
+//! write, or remove the file, for example.
+//! If the file is removed while a process is holding a lock on it,
+//! the process that holds the lock does not get any error or notification.
+//! Furthermore, you can create a new file with the same name and lock the new file,
+//! while the old process is still running.
+//! Deleting the lock file while the locking process is still running is a bad idea!
+
+use std::{fs, os::unix::prelude::AsRawFd, path::Path};
+
+use anyhow::Context;
+use nix::fcntl;
+
+use crate::crashsafe;
+
+pub enum LockCreationResult {
+    Created {
+        new_lock_contents: String,
+        file: fs::File,
+    },
+    AlreadyLocked {
+        existing_lock_contents: String,
+    },
+    CreationFailed(anyhow::Error),
+}
+
+/// Creates a lock file in the path given and writes the given contents into the file.
+/// Note: The lock is automatically released when the file closed. You might want to use Box::leak to make sure it lives until the end of the program.
+pub fn create_lock_file(lock_file_path: &Path, contents: String) -> LockCreationResult {
+    let lock_file = match fs::OpenOptions::new()
+        .create(true) // O_CREAT
+        .write(true)
+        .open(lock_file_path)
+        .context("Failed to open lock file")
+    {
+        Ok(file) => file,
+        Err(e) => return LockCreationResult::CreationFailed(e),
+    };
+
+    match fcntl::flock(
+        lock_file.as_raw_fd(),
+        fcntl::FlockArg::LockExclusiveNonblock,
+    ) {
+        Ok(()) => {
+            match lock_file
+                .set_len(0)
+                .context("Failed to truncate lockfile")
+                .and_then(|()| {
+                    fs::write(lock_file_path, &contents).with_context(|| {
+                        format!("Failed to write '{contents}' contents into lockfile")
+                    })
+                })
+                .and_then(|()| {
+                    crashsafe::fsync_file_and_parent(lock_file_path)
+                        .context("Failed to fsync lockfile")
+                }) {
+                Ok(()) => LockCreationResult::Created {
+                    new_lock_contents: contents,
+                    file: lock_file,
+                },
+                Err(e) => LockCreationResult::CreationFailed(e),
+            }
+        }
+        Err(nix::errno::Errno::EAGAIN) => {
+            match fs::read_to_string(lock_file_path).context("Failed to read lockfile contents") {
+                Ok(existing_lock_contents) => LockCreationResult::AlreadyLocked {
+                    existing_lock_contents,
+                },
+                Err(e) => LockCreationResult::CreationFailed(e),
+            }
+        }
+        Err(e) => {
+            LockCreationResult::CreationFailed(anyhow::anyhow!("Failed to lock lockfile: {e}"))
+        }
+    }
+}

libs/utils/src/logging.rs

@@ -1,19 +1,28 @@
-use std::{
-    fs::{File, OpenOptions},
-    path::Path,
-};
+use std::str::FromStr;
 
-use anyhow::{Context, Result};
+use anyhow::Context;
+use strum_macros::{EnumString, EnumVariantNames};
 
-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()
-        .create(true)
-        .append(true)
-        .open(&log_filename)
-        .with_context(|| format!("failed to open {:?}", log_filename.as_ref()))?;
+#[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_format: LogFormat) -> anyhow::Result<()> {
     let default_filter_str = "info";
 
     // We fall back to printing all spans at info-level or above if
@@ -23,20 +32,14 @@ pub fn init(log_filename: impl AsRef<Path>, daemonize: bool) -> Result<File> {
 
     let base_logger = tracing_subscriber::fmt()
         .with_env_filter(env_filter)
-        .with_target(false) // don't include event targets
-        .with_ansi(false); // don't use colors in log file;
+        .with_target(false)
+        .with_ansi(false)
+        .with_writer(std::io::stdout);
 
-    // 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();
+    match log_format {
+        LogFormat::Json => base_logger.json().init(),
+        LogFormat::Plain => base_logger.init(),
     }
 
-    Ok(log_file)
+    Ok(())
 }

libs/utils/src/lsn.rs

@@ -13,7 +13,7 @@ use crate::seqwait::MonotonicCounter;
 pub const XLOG_BLCKSZ: u32 = 8192;
 
 /// A Postgres LSN (Log Sequence Number), also known as an XLogRecPtr
-#[derive(Clone, Copy, Eq, Ord, PartialEq, PartialOrd, Serialize, Deserialize)]
+#[derive(Clone, Copy, Eq, Ord, PartialEq, PartialOrd, Hash, Serialize, Deserialize)]
 #[serde(transparent)]
 pub struct Lsn(pub u64);
 

libs/utils/src/postgres_backend.rs

@@ -3,10 +3,10 @@
 //! implementation determining how to process the queries. Currently its API
 //! is rather narrow, but we can extend it once required.
 
-use crate::pq_proto::{BeMessage, BeParameterStatusMessage, FeMessage, FeStartupPacket};
 use crate::sock_split::{BidiStream, ReadStream, WriteStream};
 use anyhow::{bail, ensure, Context, Result};
 use bytes::{Bytes, BytesMut};
+use pq_proto::{BeMessage, BeParameterStatusMessage, FeMessage, FeStartupPacket};
 use rand::Rng;
 use serde::{Deserialize, Serialize};
 use std::fmt;

libs/utils/src/postgres_backend_async.rs

@@ -4,9 +4,9 @@
 //! is rather narrow, but we can extend it once required.
 
 use crate::postgres_backend::AuthType;
-use crate::pq_proto::{BeMessage, BeParameterStatusMessage, FeMessage, FeStartupPacket};
 use anyhow::{bail, Context, Result};
 use bytes::{Bytes, BytesMut};
+use pq_proto::{BeMessage, BeParameterStatusMessage, FeMessage, FeStartupPacket};
 use rand::Rng;
 use std::future::Future;
 use std::net::SocketAddr;
@@ -15,7 +15,7 @@ use std::sync::Arc;
 use std::task::Poll;
 use tracing::{debug, error, trace};
 
-use tokio::io::{AsyncRead, AsyncWrite, AsyncWriteExt};
+use tokio::io::{AsyncRead, AsyncWrite, AsyncWriteExt, BufReader};
 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(tokio::net::TcpStream),
-    Tls(Box<tokio_rustls::server::TlsStream<tokio::net::TcpStream>>),
+    Unencrypted(BufReader<tokio::net::TcpStream>),
+    Tls(Box<tokio_rustls::server::TlsStream<BufReader<tokio::net::TcpStream>>>),
     Broken,
 }
 
@@ -157,7 +157,7 @@ impl PostgresBackend {
         let peer_addr = socket.peer_addr()?;
 
         Ok(Self {
-            stream: Stream::Unencrypted(socket),
+            stream: Stream::Unencrypted(BufReader::new(socket)),
             buf_out: BytesMut::with_capacity(10 * 1024),
             state: ProtoState::Initialization,
             md5_salt: [0u8; 4],

pageserver/Cargo.toml

@@ -12,62 +12,67 @@ testing = ["fail/failpoints"]
 profiling = ["pprof"]
 
 [dependencies]
+amplify_num = { git = "https://github.com/hlinnaka/rust-amplify.git", branch = "unsigned-int-perf" }
+anyhow = { version = "1.0", features = ["backtrace"] }
 async-stream = "0.3"
 async-trait = "0.1"
-chrono = "0.4.19"
-rand = "0.8.3"
-regex = "1.4.5"
-bytes = "1.0.1"
 byteorder = "1.4.3"
+bytes = "1.0.1"
+chrono = "0.4.19"
+clap = { version = "4.0", features = ["string"] }
+close_fds = "0.3.2"
+const_format = "0.2.21"
+crc32c = "0.6.0"
+crossbeam-utils = "0.8.5"
+fail = "0.5.0"
 futures = "0.3.13"
+git-version = "0.3.5"
 hex = "0.4.3"
+humantime = "2.1.0"
+humantime-serde = "1.1.1"
 hyper = "0.14"
 itertools = "0.10.3"
-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"] }
-postgres-types = { git = "https://github.com/neondatabase/rust-postgres.git", rev="d052ee8b86fff9897c77b0fe89ea9daba0e1fa38" }
-postgres-protocol = { git = "https://github.com/neondatabase/rust-postgres.git", rev="d052ee8b86fff9897c77b0fe89ea9daba0e1fa38" }
+nix = "0.25"
+num-traits = "0.2.15"
+once_cell = "1.13.0"
 postgres = { git = "https://github.com/neondatabase/rust-postgres.git", rev="d052ee8b86fff9897c77b0fe89ea9daba0e1fa38" }
-tokio-postgres = { git = "https://github.com/neondatabase/rust-postgres.git", rev="d052ee8b86fff9897c77b0fe89ea9daba0e1fa38" }
-anyhow = { version = "1.0", features = ["backtrace"] }
-crc32c = "0.6.0"
-thiserror = "1.0"
-tar = "0.4.33"
-humantime = "2.1.0"
+postgres-protocol = { git = "https://github.com/neondatabase/rust-postgres.git", rev="d052ee8b86fff9897c77b0fe89ea9daba0e1fa38" }
+postgres-types = { git = "https://github.com/neondatabase/rust-postgres.git", rev="d052ee8b86fff9897c77b0fe89ea9daba0e1fa38" }
+pprof = { git = "https://github.com/neondatabase/pprof-rs.git", branch = "wallclock-profiling", features = ["flamegraph"], optional = true }
+rand = "0.8.3"
+regex = "1.4.5"
+rstar = "0.9.3"
+scopeguard = "1.1.0"
 serde = { version = "1.0", features = ["derive"] }
 serde_json = "1"
 serde_with = "2.0"
-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"] }
-scopeguard = "1.1.0"
-const_format = "0.2.21"
-tracing = "0.1.36"
 signal-hook = "0.3.10"
+svg_fmt = "0.4.1"
+tar = "0.4.33"
+thiserror = "1.0"
+tokio = { version = "1.17", features = ["process", "sync", "macros", "fs", "rt", "io-util", "time"] }
+tokio-postgres = { git = "https://github.com/neondatabase/rust-postgres.git", rev="d052ee8b86fff9897c77b0fe89ea9daba0e1fa38" }
+tokio-util = { version = "0.7.3", features = ["io", "io-util"] }
+toml_edit = { version = "0.14", features = ["easy"] }
+tracing = "0.1.36"
 url = "2"
-nix = "0.25"
-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 = "0.4.1"
-
-pageserver_api = { path = "../libs/pageserver_api" }
-postgres_ffi = { path = "../libs/postgres_ffi" }
-etcd_broker = { path = "../libs/etcd_broker" }
-metrics = { path = "../libs/metrics" }
-utils = { path = "../libs/utils" }
-remote_storage = { path = "../libs/remote_storage" }
-workspace_hack = { version = "0.1", path = "../workspace_hack" }
-close_fds = "0.3.2"
 walkdir = "2.3.2"
 
+etcd_broker = { path = "../libs/etcd_broker" }
+metrics = { path = "../libs/metrics" }
+pageserver_api = { path = "../libs/pageserver_api" }
+postgres_ffi = { path = "../libs/postgres_ffi" }
+pq_proto = { path = "../libs/pq_proto" }
+remote_storage = { path = "../libs/remote_storage" }
+tenant_size_model = { path = "../libs/tenant_size_model" }
+utils = { path = "../libs/utils" }
+workspace_hack = { version = "0.1", path = "../workspace_hack" }
+
 [dev-dependencies]
+criterion = "0.4"
 hex-literal = "0.3"
 tempfile = "3.2"
+
+[[bench]]
+name = "bench_layer_map"
+harness = false