Add assertion

.dockerignore

@@ -15,10 +15,8 @@
 !proxy/
 !safekeeper/
 !storage_broker/
-!trace/
 !vendor/postgres-v14/
 !vendor/postgres-v15/
 !workspace_hack/
 !neon_local/
 !scripts/ninstall.sh
-!vm-cgconfig.conf

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

@@ -123,8 +123,8 @@ runs:
           exit 1
         fi
         if [[ "${{ inputs.run_in_parallel }}" == "true" ]]; then
-          # -n16 uses sixteen processes to run tests via pytest-xdist
-          EXTRA_PARAMS="-n16 $EXTRA_PARAMS"
+          # -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

.github/ansible/deploy.yaml

@@ -91,15 +91,6 @@
       tags:
       - pageserver
 
-    # used in `pageserver.service` template
-    - name: learn current availability_zone
-      shell:
-        cmd: "curl -s http://169.254.169.254/latest/meta-data/placement/availability-zone"
-      register: ec2_availability_zone
-
-    - set_fact: 
-        ec2_availability_zone={{ ec2_availability_zone.stdout }}
-
     - name: upload systemd service definition
       ansible.builtin.template:
         src: systemd/pageserver.service
@@ -127,7 +118,7 @@
         cmd: |
           INSTANCE_ID=$(curl -s http://169.254.169.254/latest/meta-data/instance-id)
           curl -sfS -H "Authorization: Bearer {{ CONSOLE_API_TOKEN }}" {{ console_mgmt_base_url }}/management/api/v2/pageservers/$INSTANCE_ID | jq '.version = {{ current_version }}' > /tmp/new_version
-          curl -sfS -H "Authorization: Bearer {{ CONSOLE_API_TOKEN }}" -H "Content-Type: application/json" -X POST -d@/tmp/new_version {{ console_mgmt_base_url }}/management/api/v2/pageservers
+          curl -sfS -H "Authorization: Bearer {{ CONSOLE_API_TOKEN }}" -X POST -d@/tmp/new_version {{ console_mgmt_base_url }}/management/api/v2/pageservers
       tags:
       - pageserver
 
@@ -162,15 +153,6 @@
       tags:
       - safekeeper
 
-    # used in `safekeeper.service` template
-    - name: learn current availability_zone
-      shell:
-        cmd: "curl -s http://169.254.169.254/latest/meta-data/placement/availability-zone"
-      register: ec2_availability_zone
-
-    - set_fact: 
-        ec2_availability_zone={{ ec2_availability_zone.stdout }}
-
     # in the future safekeepers should discover pageservers byself
     # but currently use first pageserver that was discovered
     - name: set first pageserver var for safekeepers
@@ -206,6 +188,6 @@
         cmd: |
           INSTANCE_ID=$(curl -s http://169.254.169.254/latest/meta-data/instance-id)
           curl -sfS -H "Authorization: Bearer {{ CONSOLE_API_TOKEN }}" {{ console_mgmt_base_url }}/management/api/v2/safekeepers/$INSTANCE_ID | jq '.version = {{ current_version }}' > /tmp/new_version
-          curl -sfS -H "Authorization: Bearer {{ CONSOLE_API_TOKEN }}" -H "Content-Type: application/json" -X POST -d@/tmp/new_version {{ console_mgmt_base_url }}/management/api/v2/safekeepers
+          curl -sfS -H "Authorization: Bearer {{ CONSOLE_API_TOKEN }}" -X POST -d@/tmp/new_version {{ console_mgmt_base_url }}/management/api/v2/safekeepers
       tags:
       - safekeeper

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

@@ -2,11 +2,11 @@ storage:
   vars:
     bucket_name: neon-prod-storage-ap-southeast-1
     bucket_region: ap-southeast-1
-    console_mgmt_base_url: http://neon-internal-api.aws.neon.tech
+    console_mgmt_base_url: http://console-release.local
     broker_endpoint: http://storage-broker-lb.epsilon.ap-southeast-1.internal.aws.neon.tech:50051
     pageserver_config_stub:
       pg_distrib_dir: /usr/local
-      metric_collection_endpoint: http://neon-internal-api.aws.neon.tech/billing/api/v1/usage_events
+      metric_collection_endpoint: http://console-release.local/billing/api/v1/usage_events
       metric_collection_interval: 10min
       remote_storage:
         bucket_name: "{{ bucket_name }}"
@@ -32,7 +32,7 @@ storage:
       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
-        safekeeper-3.ap-southeast-1.aws.neon.tech:
-          ansible_host:  i-05481f3bc88cfc2d4

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

@@ -2,11 +2,11 @@ storage:
   vars:
     bucket_name: neon-prod-storage-eu-central-1
     bucket_region: eu-central-1
-    console_mgmt_base_url: http://neon-internal-api.aws.neon.tech
+    console_mgmt_base_url: http://console-release.local
     broker_endpoint: http://storage-broker-lb.gamma.eu-central-1.internal.aws.neon.tech:50051
     pageserver_config_stub:
       pg_distrib_dir: /usr/local
-      metric_collection_endpoint: http://neon-internal-api.aws.neon.tech/billing/api/v1/usage_events
+      metric_collection_endpoint: http://console-release.local/billing/api/v1/usage_events
       metric_collection_interval: 10min
       remote_storage:
         bucket_name: "{{ bucket_name }}"
@@ -27,8 +27,6 @@ storage:
           ansible_host:  i-0cd8d316ecbb715be
         pageserver-1.eu-central-1.aws.neon.tech:
           ansible_host:  i-090044ed3d383fef0
-        pageserver-2.eu-central-1.aws.neon.tech:
-          ansible_host:  i-033584edf3f4b6742
 
     safekeepers:
       hosts:

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

@@ -2,11 +2,11 @@ storage:
   vars:
     bucket_name: neon-prod-storage-us-east-2
     bucket_region: us-east-2
-    console_mgmt_base_url: http://neon-internal-api.aws.neon.tech
+    console_mgmt_base_url: http://console-release.local
     broker_endpoint: http://storage-broker-lb.delta.us-east-2.internal.aws.neon.tech:50051
     pageserver_config_stub:
       pg_distrib_dir: /usr/local
-      metric_collection_endpoint: http://neon-internal-api.aws.neon.tech/billing/api/v1/usage_events
+      metric_collection_endpoint: http://console-release.local/billing/api/v1/usage_events
       metric_collection_interval: 10min
       remote_storage:
         bucket_name: "{{ bucket_name }}"
@@ -27,8 +27,6 @@ storage:
           ansible_host:  i-062227ba7f119eb8c
         pageserver-1.us-east-2.aws.neon.tech:
           ansible_host:  i-0b3ec0afab5968938
-        pageserver-2.us-east-2.aws.neon.tech:
-          ansible_host:  i-0d7a1c4325e71421d
 
     safekeepers:
       hosts:

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

@@ -2,11 +2,11 @@ storage:
   vars:
     bucket_name: neon-prod-storage-us-west-2
     bucket_region: us-west-2
-    console_mgmt_base_url: http://neon-internal-api.aws.neon.tech
+    console_mgmt_base_url: http://console-release.local
     broker_endpoint: http://storage-broker-lb.eta.us-west-2.internal.aws.neon.tech:50051
     pageserver_config_stub:
       pg_distrib_dir: /usr/local
-      metric_collection_endpoint: http://neon-internal-api.aws.neon.tech/billing/api/v1/usage_events
+      metric_collection_endpoint: http://console-release.local/billing/api/v1/usage_events
       metric_collection_interval: 10min
       remote_storage:
         bucket_name: "{{ bucket_name }}"
@@ -29,8 +29,6 @@ storage:
           ansible_host: i-0c834be1dddba8b3f
         pageserver-2.us-west-2.aws.neon.tech:
           ansible_host: i-051642d372c0a4f32
-        pageserver-3.us-west-2.aws.neon.tech:
-          ansible_host: i-00c3844beb9ad1c6b
 
     safekeepers:
       hosts:

.github/ansible/production.hosts.yaml

@@ -0,0 +1,40 @@
+---
+storage:
+  vars:
+    console_mgmt_base_url: http://console-release.local
+    bucket_name: zenith-storage-oregon
+    bucket_region: us-west-2
+    broker_endpoint: http://storage-broker.prod.local:50051
+    pageserver_config_stub:
+      pg_distrib_dir: /usr/local
+      metric_collection_endpoint: http://console-release.local/billing/api/v1/usage_events
+      metric_collection_interval: 10min
+      remote_storage:
+        bucket_name: "{{ bucket_name }}"
+        bucket_region: "{{ bucket_region }}"
+        prefix_in_bucket: "{{ inventory_hostname }}"
+    safekeeper_s3_prefix: prod-1/wal
+    hostname_suffix: ".local"
+    remote_user: admin
+    sentry_environment: production
+
+  children:
+    pageservers:
+      hosts:
+        zenith-1-ps-2:
+          console_region_id: aws-us-west-2
+        zenith-1-ps-3:
+          console_region_id: aws-us-west-2
+        zenith-1-ps-4:
+          console_region_id: aws-us-west-2
+        zenith-1-ps-5:
+          console_region_id: aws-us-west-2
+
+    safekeepers:
+      hosts:
+        zenith-1-sk-1:
+          console_region_id: aws-us-west-2
+        zenith-1-sk-2:
+          console_region_id: aws-us-west-2
+        zenith-1-sk-4:
+          console_region_id: aws-us-west-2

.github/ansible/scripts/init_pageserver.sh

@@ -26,7 +26,7 @@ EOF
 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=$(curl -sf -X POST -H "Authorization: Bearer {{ CONSOLE_API_TOKEN }}" -H "Content-Type: application/json" {{ console_mgmt_base_url }}/management/api/v2/pageservers -d@/tmp/payload | jq -r '.id')
+    ID=$(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

@@ -25,7 +25,7 @@ EOF
 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=$(curl -sf -X POST -H "Authorization: Bearer {{ CONSOLE_API_TOKEN }}" -H "Content-Type: application/json" {{ console_mgmt_base_url }}/management/api/v2/safekeepers -d@/tmp/payload | jq -r '.id')
+    ID=$(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/staging.eu-west-1.hosts.yaml

@@ -2,17 +2,12 @@ storage:
   vars:
     bucket_name: neon-dev-storage-eu-west-1
     bucket_region: eu-west-1
-    console_mgmt_base_url: http://neon-internal-api.aws.neon.build
+    console_mgmt_base_url: http://console-staging.local
     broker_endpoint: http://storage-broker-lb.zeta.eu-west-1.internal.aws.neon.build:50051
     pageserver_config_stub:
       pg_distrib_dir: /usr/local
-      metric_collection_endpoint: http://neon-internal-api.aws.neon.build/billing/api/v1/usage_events
+      metric_collection_endpoint: http://console-staging.local/billing/api/v1/usage_events
       metric_collection_interval: 10min
-      tenant_config:
-        eviction_policy:
-          kind: "LayerAccessThreshold"
-          period: "20m"
-          threshold: "20m"
       remote_storage:
         bucket_name: "{{ bucket_name }}"
         bucket_region: "{{ bucket_region }}"

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

@@ -2,17 +2,12 @@ storage:
   vars:
     bucket_name: neon-staging-storage-us-east-2
     bucket_region: us-east-2
-    console_mgmt_base_url: http://neon-internal-api.aws.neon.build
+    console_mgmt_base_url: http://console-staging.local
     broker_endpoint: http://storage-broker-lb.beta.us-east-2.internal.aws.neon.build:50051
     pageserver_config_stub:
       pg_distrib_dir: /usr/local
-      metric_collection_endpoint: http://neon-internal-api.aws.neon.build/billing/api/v1/usage_events
+      metric_collection_endpoint: http://console-staging.local/billing/api/v1/usage_events
       metric_collection_interval: 10min
-      tenant_config:
-        eviction_policy:
-          kind: "LayerAccessThreshold"
-          period: "20m"
-          threshold: "20m"
       remote_storage:
         bucket_name: "{{ bucket_name }}"
         bucket_region: "{{ bucket_region }}"
@@ -36,8 +31,6 @@ storage:
           ansible_host: i-01e31cdf7e970586a
         pageserver-3.us-east-2.aws.neon.build:
           ansible_host: i-0602a0291365ef7cc
-        pageserver-99.us-east-2.aws.neon.build:
-          ansible_host: i-0c39491109bb88824
 
     safekeepers:
       hosts:
@@ -47,5 +40,3 @@ storage:
           ansible_host: i-0171efc3604a7b907
         safekeeper-2.us-east-2.aws.neon.build:
           ansible_host: i-0de0b03a51676a6ce
-        safekeeper-99.us-east-2.aws.neon.build:
-          ansible_host: i-0d61b6a2ea32028d5

.github/ansible/systemd/pageserver.service

@@ -6,7 +6,7 @@ After=network.target auditd.service
 Type=simple
 User=pageserver
 Environment=RUST_BACKTRACE=1 NEON_REPO_DIR=/storage/pageserver LD_LIBRARY_PATH=/usr/local/v14/lib SENTRY_DSN={{ SENTRY_URL_PAGESERVER }} SENTRY_ENVIRONMENT={{ sentry_environment }}
-ExecStart=/usr/local/bin/pageserver -c "pg_distrib_dir='/usr/local'" -c "listen_pg_addr='0.0.0.0:6400'" -c "listen_http_addr='0.0.0.0:9898'" -c "broker_endpoint='{{ broker_endpoint }}'" -c "availability_zone='{{ ec2_availability_zone }}'" -D /storage/pageserver/data
+ExecStart=/usr/local/bin/pageserver -c "pg_distrib_dir='/usr/local'" -c "listen_pg_addr='0.0.0.0:6400'" -c "listen_http_addr='0.0.0.0:9898'" -c "broker_endpoint='{{ broker_endpoint }}'" -D /storage/pageserver/data
 ExecReload=/bin/kill -HUP $MAINPID
 KillMode=mixed
 KillSignal=SIGINT

.github/ansible/systemd/safekeeper.service

@@ -6,7 +6,7 @@ After=network.target auditd.service
 Type=simple
 User=safekeeper
 Environment=RUST_BACKTRACE=1 NEON_REPO_DIR=/storage/safekeeper/data LD_LIBRARY_PATH=/usr/local/v14/lib SENTRY_DSN={{ SENTRY_URL_SAFEKEEPER }} SENTRY_ENVIRONMENT={{ sentry_environment }}
-ExecStart=/usr/local/bin/safekeeper -l {{ inventory_hostname }}{{ hostname_suffix }}:6500 --listen-http {{ inventory_hostname }}{{ hostname_suffix }}:7676 -D /storage/safekeeper/data --broker-endpoint={{ broker_endpoint }} --remote-storage='{bucket_name="{{bucket_name}}", bucket_region="{{bucket_region}}", prefix_in_bucket="{{ safekeeper_s3_prefix }}"}' --availability-zone={{ ec2_availability_zone }}
+ExecStart=/usr/local/bin/safekeeper -l {{ inventory_hostname }}{{ hostname_suffix }}:6500 --listen-http {{ inventory_hostname }}{{ hostname_suffix }}:7676 -D /storage/safekeeper/data --broker-endpoint={{ broker_endpoint }} --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-eu-west-1-zeta.neon-proxy-scram.yaml

@@ -1,31 +1,16 @@
 # Helm chart values for neon-proxy-scram.
 # This is a YAML-formatted file.
 
-deploymentStrategy:
-  type: RollingUpdate
-  rollingUpdate:
-    maxSurge: 100%
-    maxUnavailable: 50%
-
-# Delay the kill signal by 7 days (7 * 24 * 60 * 60)
-# The pod(s) will stay in Terminating, keeps the existing connections
-# but doesn't receive new ones
-containerLifecycle:
-  preStop:
-    exec:
-      command: ["/bin/sh", "-c", "sleep 604800"]
-terminationGracePeriodSeconds: 604800
-
 image:
   repository: neondatabase/neon
 
 settings:
   authBackend: "console"
-  authEndpoint: "http://neon-internal-api.aws.neon.build/management/api/v2"
+  authEndpoint: "http://console-staging.local/management/api/v2"
   domain: "*.eu-west-1.aws.neon.build"
   sentryEnvironment: "staging"
   wssPort: 8443
-  metricCollectionEndpoint: "http://neon-internal-api.aws.neon.build/billing/api/v1/usage_events"
+  metricCollectionEndpoint: "http://console-staging.local/billing/api/v1/usage_events"
   metricCollectionInterval: "1min"
 
 # -- Additional labels for neon-proxy pods

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

@@ -8,9 +8,8 @@ settings:
   authBackend: "link"
   authEndpoint: "https://console.stage.neon.tech/authenticate_proxy_request/"
   uri: "https://console.stage.neon.tech/psql_session/"
-  domain: "pg.neon.build"
   sentryEnvironment: "staging"
-  metricCollectionEndpoint: "http://neon-internal-api.aws.neon.build/billing/api/v1/usage_events"
+  metricCollectionEndpoint: "http://console-staging.local/billing/api/v1/usage_events"
   metricCollectionInterval: "1min"
 
 # -- Additional labels for neon-proxy-link pods

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

@@ -6,11 +6,11 @@ image:
 
 settings:
   authBackend: "console"
-  authEndpoint: "http://neon-internal-api.aws.neon.build/management/api/v2"
+  authEndpoint: "http://console-staging.local/management/api/v2"
   domain: "*.cloud.stage.neon.tech"
   sentryEnvironment: "staging"
   wssPort: 8443
-  metricCollectionEndpoint: "http://neon-internal-api.aws.neon.build/billing/api/v1/usage_events"
+  metricCollectionEndpoint: "http://console-staging.local/billing/api/v1/usage_events"
   metricCollectionInterval: "1min"
 
 # -- Additional labels for neon-proxy pods

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

@@ -1,31 +1,16 @@
 # Helm chart values for neon-proxy-scram.
 # This is a YAML-formatted file.
 
-deploymentStrategy:
-  type: RollingUpdate
-  rollingUpdate:
-    maxSurge: 100%
-    maxUnavailable: 50%
-
-# Delay the kill signal by 7 days (7 * 24 * 60 * 60)
-# The pod(s) will stay in Terminating, keeps the existing connections
-# but doesn't receive new ones
-containerLifecycle:
-  preStop:
-    exec:
-      command: ["/bin/sh", "-c", "sleep 604800"]
-terminationGracePeriodSeconds: 604800
-
 image:
   repository: neondatabase/neon
 
 settings:
   authBackend: "console"
-  authEndpoint: "http://neon-internal-api.aws.neon.build/management/api/v2"
+  authEndpoint: "http://console-staging.local/management/api/v2"
   domain: "*.us-east-2.aws.neon.build"
   sentryEnvironment: "staging"
   wssPort: 8443
-  metricCollectionEndpoint: "http://neon-internal-api.aws.neon.build/billing/api/v1/usage_events"
+  metricCollectionEndpoint: "http://console-staging.local/billing/api/v1/usage_events"
   metricCollectionInterval: "1min"
 
 # -- Additional labels for neon-proxy pods

.github/helm-values/prod-ap-southeast-1-epsilon.neon-proxy-scram.yaml

@@ -1,32 +1,16 @@
 # Helm chart values for neon-proxy-scram.
 # This is a YAML-formatted file.
 
-deploymentStrategy:
-  type: RollingUpdate
-  rollingUpdate:
-    maxSurge: 100%
-    maxUnavailable: 50%
-
-# Delay the kill signal by 7 days (7 * 24 * 60 * 60)
-# The pod(s) will stay in Terminating, keeps the existing connections
-# but doesn't receive new ones
-containerLifecycle:
-  preStop:
-    exec:
-      command: ["/bin/sh", "-c", "sleep 604800"]
-terminationGracePeriodSeconds: 604800
-
-
 image:
   repository: neondatabase/neon
 
 settings:
   authBackend: "console"
-  authEndpoint: "http://neon-internal-api.aws.neon.tech/management/api/v2"
+  authEndpoint: "http://console-release.local/management/api/v2"
   domain: "*.ap-southeast-1.aws.neon.tech"
   sentryEnvironment: "production"
   wssPort: 8443
-  metricCollectionEndpoint: "http://neon-internal-api.aws.neon.tech/billing/api/v1/usage_events"
+  metricCollectionEndpoint: "http://console-release.local/billing/api/v1/usage_events"
   metricCollectionInterval: "10min"
 
 # -- Additional labels for neon-proxy pods

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

@@ -1,32 +1,16 @@
 # Helm chart values for neon-proxy-scram.
 # This is a YAML-formatted file.
 
-deploymentStrategy:
-  type: RollingUpdate
-  rollingUpdate:
-    maxSurge: 100%
-    maxUnavailable: 50%
-
-# Delay the kill signal by 7 days (7 * 24 * 60 * 60)
-# The pod(s) will stay in Terminating, keeps the existing connections
-# but doesn't receive new ones
-containerLifecycle:
-  preStop:
-    exec:
-      command: ["/bin/sh", "-c", "sleep 604800"]
-terminationGracePeriodSeconds: 604800
-
-
 image:
   repository: neondatabase/neon
 
 settings:
   authBackend: "console"
-  authEndpoint: "http://neon-internal-api.aws.neon.tech/management/api/v2"
+  authEndpoint: "http://console-release.local/management/api/v2"
   domain: "*.eu-central-1.aws.neon.tech"
   sentryEnvironment: "production"
   wssPort: 8443
-  metricCollectionEndpoint: "http://neon-internal-api.aws.neon.tech/billing/api/v1/usage_events"
+  metricCollectionEndpoint: "http://console-release.local/billing/api/v1/usage_events"
   metricCollectionInterval: "10min"
 
 # -- Additional labels for neon-proxy pods

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

@@ -1,32 +1,16 @@
 # Helm chart values for neon-proxy-scram.
 # This is a YAML-formatted file.
 
-deploymentStrategy:
-  type: RollingUpdate
-  rollingUpdate:
-    maxSurge: 100%
-    maxUnavailable: 50%
-
-# Delay the kill signal by 7 days (7 * 24 * 60 * 60)
-# The pod(s) will stay in Terminating, keeps the existing connections
-# but doesn't receive new ones
-containerLifecycle:
-  preStop:
-    exec:
-      command: ["/bin/sh", "-c", "sleep 604800"]
-terminationGracePeriodSeconds: 604800
-
-
 image:
   repository: neondatabase/neon
 
 settings:
   authBackend: "console"
-  authEndpoint: "http://neon-internal-api.aws.neon.tech/management/api/v2"
+  authEndpoint: "http://console-release.local/management/api/v2"
   domain: "*.us-east-2.aws.neon.tech"
   sentryEnvironment: "production"
   wssPort: 8443
-  metricCollectionEndpoint: "http://neon-internal-api.aws.neon.tech/billing/api/v1/usage_events"
+  metricCollectionEndpoint: "http://console-release.local/billing/api/v1/usage_events"
   metricCollectionInterval: "10min"
 
 # -- Additional labels for neon-proxy pods

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

@@ -1,32 +1,16 @@
 # Helm chart values for neon-proxy-scram.
 # This is a YAML-formatted file.
 
-deploymentStrategy:
-  type: RollingUpdate
-  rollingUpdate:
-    maxSurge: 100%
-    maxUnavailable: 50%
-
-# Delay the kill signal by 7 days (7 * 24 * 60 * 60)
-# The pod(s) will stay in Terminating, keeps the existing connections
-# but doesn't receive new ones
-containerLifecycle:
-  preStop:
-    exec:
-      command: ["/bin/sh", "-c", "sleep 604800"]
-terminationGracePeriodSeconds: 604800
-
-
 image:
   repository: neondatabase/neon
 
 settings:
   authBackend: "console"
-  authEndpoint: "http://neon-internal-api.aws.neon.tech/management/api/v2"
+  authEndpoint: "http://console-release.local/management/api/v2"
   domain: "*.us-west-2.aws.neon.tech"
   sentryEnvironment: "production"
   wssPort: 8443
-  metricCollectionEndpoint: "http://neon-internal-api.aws.neon.tech/billing/api/v1/usage_events"
+  metricCollectionEndpoint: "http://console-release.local/billing/api/v1/usage_events"
   metricCollectionInterval: "10min"
 
 # -- Additional labels for neon-proxy pods

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

@@ -0,0 +1,56 @@
+# Helm chart values for neon-storage-broker
+podLabels:
+  neon_env: production
+  neon_service: storage-broker
+
+# Use L4 LB
+service:
+  # service.annotations -- Annotations to add to the service
+  annotations:
+    service.beta.kubernetes.io/aws-load-balancer-type: external  # use newer AWS Load Balancer Controller
+    service.beta.kubernetes.io/aws-load-balancer-nlb-target-type: ip
+    service.beta.kubernetes.io/aws-load-balancer-scheme: internal  # deploy LB to private subnet
+    # assign service to this name at external-dns
+    external-dns.alpha.kubernetes.io/hostname: storage-broker.prod.local
+  # service.type -- Service type
+  type: LoadBalancer
+  # service.port -- broker listen port
+  port: 50051
+
+ingress:
+  enabled: false
+
+metrics:
+  enabled: true
+  serviceMonitor:
+    enabled: true
+    selector:
+      release: kube-prometheus-stack
+
+extraManifests:
+  - apiVersion: operator.victoriametrics.com/v1beta1
+    kind: VMServiceScrape
+    metadata:
+      name: "{{ include \"neon-storage-broker.fullname\" . }}"
+      labels:
+        helm.sh/chart: neon-storage-broker-{{ .Chart.Version }}
+        app.kubernetes.io/name: neon-storage-broker
+        app.kubernetes.io/instance: neon-storage-broker
+        app.kubernetes.io/version: "{{ .Chart.AppVersion }}"
+        app.kubernetes.io/managed-by: Helm
+      namespace: "{{ .Release.Namespace }}"
+    spec:
+      selector:
+        matchLabels:
+          app.kubernetes.io/name: "neon-storage-broker"
+      endpoints:
+        - port: broker
+          path: /metrics
+          interval: 10s
+          scrapeTimeout: 10s
+      namespaceSelector:
+        matchNames:
+          - "{{ .Release.Namespace }}"
+
+settings:
+  sentryEnvironment: "production"

.github/helm-values/production.proxy-scram.yaml

@@ -1,55 +1,32 @@
-# Helm chart values for neon-proxy-scram.
-# This is a YAML-formatted file.
-
-deploymentStrategy:
-  type: RollingUpdate
-  rollingUpdate:
-    maxSurge: 100%
-    maxUnavailable: 50%
-
-# Delay the kill signal by 7 days (7 * 24 * 60 * 60)
-# The pod(s) will stay in Terminating, keeps the existing connections
-# but doesn't receive new ones
-containerLifecycle:
-  preStop:
-    exec:
-      command: ["/bin/sh", "-c", "sleep 604800"]
-terminationGracePeriodSeconds: 604800
-
-
-image:
-  repository: neondatabase/neon
-
 settings:
   authBackend: "console"
-  authEndpoint: "http://neon-internal-api.aws.neon.tech/management/api/v2"
+  authEndpoint: "http://console-release.local/management/api/v2"
   domain: "*.cloud.neon.tech"
   sentryEnvironment: "production"
   wssPort: 8443
-  metricCollectionEndpoint: "http://neon-internal-api.aws.neon.tech/billing/api/v1/usage_events"
+  metricCollectionEndpoint: "http://console-release.local/billing/api/v1/usage_events"
   metricCollectionInterval: "10min"
 
-# -- Additional labels for neon-proxy pods
 podLabels:
   zenith_service: proxy-scram
-  zenith_env: prod
+  zenith_env: production
   zenith_region: us-west-2
-  zenith_region_slug: us-west-2
+  zenith_region_slug: oregon
 
 exposedService:
   annotations:
     service.beta.kubernetes.io/aws-load-balancer-type: external
     service.beta.kubernetes.io/aws-load-balancer-nlb-target-type: ip
     service.beta.kubernetes.io/aws-load-balancer-scheme: internet-facing
-    external-dns.alpha.kubernetes.io/hostname: neon-proxy-scram-legacy.eta.us-west-2.aws.neon.tech
+    external-dns.alpha.kubernetes.io/hostname: '*.cloud.neon.tech'
   httpsPort: 443
 
-#metrics:
-#  enabled: true
-#  serviceMonitor:
-#    enabled: true
-#    selector:
-#      release: kube-prometheus-stack
+metrics:
+  enabled: true
+  serviceMonitor:
+    enabled: true
+    selector:
+      release: kube-prometheus-stack
 
 extraManifests:
   - apiVersion: operator.victoriametrics.com/v1beta1

.github/helm-values/production.proxy.yaml

@@ -1,37 +1,37 @@
-# Helm chart values for neon-proxy-link.
-# This is a YAML-formatted file.
-
-image:
-  repository: neondatabase/neon
-
 settings:
   authBackend: "link"
   authEndpoint: "https://console.neon.tech/authenticate_proxy_request/"
   uri: "https://console.neon.tech/psql_session/"
-  domain: "pg.neon.tech"
   sentryEnvironment: "production"
 
 # -- Additional labels for zenith-proxy pods
 podLabels:
   zenith_service: proxy
   zenith_env: production
-  zenith_region: us-east-2
-  zenith_region_slug: us-east-2
+  zenith_region: us-west-2
+  zenith_region_slug: oregon
 
 service:
-  type: LoadBalancer
   annotations:
     service.beta.kubernetes.io/aws-load-balancer-type: external
     service.beta.kubernetes.io/aws-load-balancer-nlb-target-type: ip
     service.beta.kubernetes.io/aws-load-balancer-scheme: internal
-    external-dns.alpha.kubernetes.io/hostname: neon-proxy-link-mgmt.delta.us-east-2.aws.neon.tech
+    external-dns.alpha.kubernetes.io/hostname: proxy-release.local
+  type: LoadBalancer
 
 exposedService:
   annotations:
     service.beta.kubernetes.io/aws-load-balancer-type: external
     service.beta.kubernetes.io/aws-load-balancer-nlb-target-type: ip
     service.beta.kubernetes.io/aws-load-balancer-scheme: internet-facing
-    external-dns.alpha.kubernetes.io/hostname: neon-proxy-link.delta.us-east-2.aws.neon.tech
+    external-dns.alpha.kubernetes.io/hostname: connect.neon.tech,pg.neon.tech
+
+metrics:
+  enabled: true
+  serviceMonitor:
+    enabled: true
+    selector:
+      release: kube-prometheus-stack
 
 extraManifests:
   - apiVersion: operator.victoriametrics.com/v1beta1

.github/workflows/build_and_test.yml

@@ -1,4 +1,4 @@
-name: Build and Test
+name: Test and Deploy
 
 on:
   push:
@@ -19,12 +19,10 @@ concurrency:
 env:
   RUST_BACKTRACE: 1
   COPT: '-Werror'
-  AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_DEV }}
-  AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_KEY_DEV }}
 
 jobs:
   tag:
-    runs-on: [ self-hosted, gen3, small ]
+    runs-on: [ self-hosted, dev, x64 ]
     container: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/base:pinned
     outputs:
       build-tag: ${{steps.build-tag.outputs.tag}}
@@ -52,9 +50,9 @@ jobs:
         id: build-tag
 
   check-codestyle-python:
-    runs-on: [ self-hosted, gen3, small ]
+    runs-on: [ self-hosted, dev, x64 ]
     container:
-      image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned
+      image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/cloud:pinned
       options: --init
 
     steps:
@@ -74,17 +72,20 @@ jobs:
       - name: Install Python deps
         run: ./scripts/pysync
 
-      - name: Run ruff to ensure code format
-        run: poetry run ruff .
+      - name: Run isort to ensure code format
+        run: poetry run isort --diff --check .
 
       - name: Run black to ensure code format
         run: poetry run black --diff --check .
 
+      - name: Run flake8 to ensure code format
+        run: poetry run flake8 .
+
       - name: Run mypy to check types
         run: poetry run mypy .
 
   check-codestyle-rust:
-    runs-on: [ self-hosted, gen3, large ]
+    runs-on: [ self-hosted, dev, x64 ]
     container:
       image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned
       options: --init
@@ -96,16 +97,16 @@ jobs:
           submodules: true
           fetch-depth: 1
 
-#      Disabled for now
-#      - name: Restore cargo deps cache
-#        id: cache_cargo
-#        uses: actions/cache@v3
-#        with:
-#          path: |
-#            !~/.cargo/registry/src
-#            ~/.cargo/git/
-#            target/
-#          key: v1-${{ runner.os }}-cargo-clippy-${{ hashFiles('rust-toolchain.toml') }}-${{ hashFiles('Cargo.lock') }}
+      - name: Restore cargo deps cache
+        id: cache_cargo
+        uses: actions/cache@v3
+        with:
+          path: |
+            ~/.cargo/registry/
+            !~/.cargo/registry/src
+            ~/.cargo/git/
+            target/
+          key: v1-${{ runner.os }}-cargo-clippy-${{ hashFiles('rust-toolchain.toml') }}-${{ hashFiles('Cargo.lock') }}
 
       # Some of our rust modules use FFI and need those to be checked
       - name: Get postgres headers
@@ -132,7 +133,7 @@ jobs:
         run: cargo deny check
 
   build-neon:
-    runs-on: [ self-hosted, gen3, large ]
+    runs-on: [ self-hosted, dev, x64 ]
     container:
       image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned
       options: --init
@@ -140,6 +141,7 @@ jobs:
       fail-fast: false
       matrix:
         build_type: [ debug, release ]
+
     env:
       BUILD_TYPE: ${{ matrix.build_type }}
       GIT_VERSION: ${{ github.sha }}
@@ -192,26 +194,24 @@ jobs:
           echo "cov_prefix=${cov_prefix}" >> $GITHUB_ENV
           echo "CARGO_FEATURES=${CARGO_FEATURES}" >> $GITHUB_ENV
           echo "CARGO_FLAGS=${CARGO_FLAGS}" >> $GITHUB_ENV
-          echo "CARGO_HOME=${GITHUB_WORKSPACE}/.cargo" >> $GITHUB_ENV
 
-      # Disabled for now
       # Don't include the ~/.cargo/registry/src directory. It contains just
       # uncompressed versions of the crates in ~/.cargo/registry/cache
       # directory, and it's faster to let 'cargo' to rebuild it from the
       # compressed crates.
-#      - name: Cache cargo deps
-#        id: cache_cargo
-#        uses: actions/cache@v3
-#        with:
-#          path: |
-#            ~/.cargo/registry/
-#            !~/.cargo/registry/src
-#            ~/.cargo/git/
-#            target/
-#          # Fall back to older versions of the key, if no cache for current Cargo.lock was found
-#          key: |
-#            v1-${{ runner.os }}-${{ matrix.build_type }}-cargo-${{ hashFiles('rust-toolchain.toml') }}-${{ hashFiles('Cargo.lock') }}
-#            v1-${{ runner.os }}-${{ matrix.build_type }}-cargo-${{ hashFiles('rust-toolchain.toml') }}-
+      - name: Cache cargo deps
+        id: cache_cargo
+        uses: actions/cache@v3
+        with:
+          path: |
+            ~/.cargo/registry/
+            !~/.cargo/registry/src
+            ~/.cargo/git/
+            target/
+          # Fall back to older versions of the key, if no cache for current Cargo.lock was found
+          key: |
+            v1-${{ runner.os }}-${{ matrix.build_type }}-cargo-${{ hashFiles('rust-toolchain.toml') }}-${{ hashFiles('Cargo.lock') }}
+            v1-${{ runner.os }}-${{ matrix.build_type }}-cargo-${{ hashFiles('rust-toolchain.toml') }}-
 
       - name: Cache postgres v14 build
         id: cache_pg_14
@@ -301,7 +301,7 @@ jobs:
         uses: ./.github/actions/save-coverage-data
 
   regress-tests:
-    runs-on: [ self-hosted, gen3, large ]
+    runs-on: [ self-hosted, dev, x64 ]
     container:
       image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned
       options: --init
@@ -334,7 +334,7 @@ jobs:
         uses: ./.github/actions/save-coverage-data
 
   benchmarks:
-    runs-on: [ self-hosted, gen3, small ]
+    runs-on: [ self-hosted, dev, x64 ]
     container:
       image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned
       options: --init
@@ -365,7 +365,7 @@ jobs:
       # while coverage is currently collected for the debug ones
 
   merge-allure-report:
-    runs-on: [ self-hosted, gen3, small ]
+    runs-on: [ self-hosted, dev, x64 ]
     container:
       image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned
       options: --init
@@ -402,7 +402,7 @@ jobs:
           DATABASE_URL="$TEST_RESULT_CONNSTR" poetry run python3 scripts/ingest_regress_test_result.py --revision ${SHA} --reference ${GITHUB_REF} --build-type ${BUILD_TYPE} --ingest suites.json
 
   coverage-report:
-    runs-on: [ self-hosted, gen3, small ]
+    runs-on: [ self-hosted, dev, x64 ]
     container:
       image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned
       options: --init
@@ -418,17 +418,16 @@ jobs:
           submodules: true
           fetch-depth: 1
 
-#      Disabled for now
-#      - name: Restore cargo deps cache
-#        id: cache_cargo
-#        uses: actions/cache@v3
-#        with:
-#          path: |
-#            ~/.cargo/registry/
-#            !~/.cargo/registry/src
-#            ~/.cargo/git/
-#            target/
-#          key: v1-${{ runner.os }}-${{ matrix.build_type }}-cargo-${{ hashFiles('rust-toolchain.toml') }}-${{ hashFiles('Cargo.lock') }}
+      - name: Restore cargo deps cache
+        id: cache_cargo
+        uses: actions/cache@v3
+        with:
+          path: |
+            ~/.cargo/registry/
+            !~/.cargo/registry/src
+            ~/.cargo/git/
+            target/
+          key: v1-${{ runner.os }}-${{ matrix.build_type }}-cargo-${{ hashFiles('rust-toolchain.toml') }}-${{ hashFiles('Cargo.lock') }}
 
       - name: Get Neon artifact
         uses: ./.github/actions/download
@@ -478,7 +477,7 @@ jobs:
             }"
 
   trigger-e2e-tests:
-    runs-on: [ self-hosted, gen3, small ]
+    runs-on: [ self-hosted, dev, x64 ]
     container:
       image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/base:pinned
       options: --init
@@ -523,10 +522,9 @@ jobs:
             }"
 
   neon-image:
-    runs-on: [ self-hosted, gen3, large ]
+    runs-on: [ self-hosted, dev, x64 ]
     needs: [ tag ]
-    # https://github.com/GoogleContainerTools/kaniko/issues/2005
-    container: gcr.io/kaniko-project/executor:v1.7.0-debug
+    container: gcr.io/kaniko-project/executor:v1.9.0-debug
     defaults:
       run:
         shell: sh -eu {0}
@@ -542,58 +540,12 @@ jobs:
         run: echo "{\"credsStore\":\"ecr-login\"}" > /kaniko/.docker/config.json
 
       - name: Kaniko build neon
-        run: /kaniko/executor --reproducible --snapshotMode=redo --skip-unused-stages --cache=true --cache-repo 369495373322.dkr.ecr.eu-central-1.amazonaws.com/cache --context . --build-arg GIT_VERSION=${{ github.sha }} --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/neon:${{needs.tag.outputs.build-tag}}
-
-      # Cleanup script fails otherwise - rm: cannot remove '/nvme/actions-runner/_work/_temp/_github_home/.ecr': Permission denied
-      - name: Cleanup ECR folder
-        run: rm -rf ~/.ecr
-
-
-  neon-image-depot:
-    # For testing this will run side-by-side for a few merges.
-    # This action is not really optimized yet, but gets the job done
-    runs-on: [ self-hosted, gen3, large ]
-    needs: [ tag ]
-    container: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/base:pinned
-    permissions:
-      contents: read
-      id-token: write
-
-    steps:
-      - name: Checkout
-        uses: actions/checkout@v3
-        with:
-          submodules: true
-          fetch-depth: 0
-
-      - name: Setup go
-        uses: actions/setup-go@v3
-        with:
-          go-version: '1.19'
-
-      - name: Set up Depot CLI
-        uses: depot/setup-action@v1
-
-      - name: Install Crane & ECR helper
-        run: go install github.com/awslabs/amazon-ecr-credential-helper/ecr-login/cli/docker-credential-ecr-login@69c85dc22db6511932bbf119e1a0cc5c90c69a7f # v0.6.0
-
-      - name: Configure ECR login
-        run: |
-          mkdir /github/home/.docker/
-          echo "{\"credsStore\":\"ecr-login\"}" > /github/home/.docker/config.json
-
-      - name: Build and push
-        uses: depot/build-push-action@v1
-        with:
-          # if no depot.json file is at the root of your repo, you must specify the project id
-          project: nrdv0s4kcs
-          push: true
-          tags: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/neon:depot-${{needs.tag.outputs.build-tag}}
+        run: /kaniko/executor --snapshotMode=redo --cache=true --cache-repo 369495373322.dkr.ecr.eu-central-1.amazonaws.com/cache --snapshotMode=redo --context . --build-arg GIT_VERSION=${{ github.sha }} --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/neon:${{needs.tag.outputs.build-tag}}
 
   compute-tools-image:
-    runs-on: [ self-hosted, gen3, large ]
+    runs-on: [ self-hosted, dev, x64 ]
     needs: [ tag ]
-    container: gcr.io/kaniko-project/executor:v1.7.0-debug
+    container: gcr.io/kaniko-project/executor:v1.9.0-debug
     defaults:
       run:
         shell: sh -eu {0}
@@ -606,14 +558,11 @@ jobs:
         run: echo "{\"credsStore\":\"ecr-login\"}" > /kaniko/.docker/config.json
 
       - name: Kaniko build compute tools
-        run: /kaniko/executor --reproducible --snapshotMode=redo --skip-unused-stages --cache=true --cache-repo 369495373322.dkr.ecr.eu-central-1.amazonaws.com/cache --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}}
-
-      - name: Cleanup ECR folder
-        run: rm -rf ~/.ecr
+        run: /kaniko/executor --snapshotMode=redo --cache=true --cache-repo 369495373322.dkr.ecr.eu-central-1.amazonaws.com/cache --snapshotMode=redo --context . --build-arg GIT_VERSION=${{ github.sha }} --dockerfile Dockerfile.compute-tools --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-tools:${{needs.tag.outputs.build-tag}}
 
   compute-node-image:
-    runs-on: [ self-hosted, gen3, large ]
-    container: gcr.io/kaniko-project/executor:v1.7.0-debug
+    runs-on: [ self-hosted, dev, x64 ]
+    container: gcr.io/kaniko-project/executor:v1.9.0-debug
     needs: [ tag ]
     strategy:
       fail-fast: false
@@ -634,13 +583,10 @@ jobs:
         run: echo "{\"credsStore\":\"ecr-login\"}" > /kaniko/.docker/config.json
 
       - name: Kaniko build compute node with extensions
-        run: /kaniko/executor --reproducible --snapshotMode=redo --skip-unused-stages --cache=true --cache-repo 369495373322.dkr.ecr.eu-central-1.amazonaws.com/cache  --context . --build-arg GIT_VERSION=${{ github.sha }} --build-arg PG_VERSION=${{ matrix.version }} --dockerfile Dockerfile.compute-node --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-${{ matrix.version }}:${{needs.tag.outputs.build-tag}}
-
-      - name: Cleanup ECR folder
-        run: rm -rf ~/.ecr
+        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=${{ github.sha }} --dockerfile Dockerfile.compute-node-${{ matrix.version }} --destination 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-${{ matrix.version }}:${{needs.tag.outputs.build-tag}}
 
   vm-compute-node-image:
-    runs-on: [ self-hosted, gen3, large ]
+    runs-on: [ self-hosted, dev, x64 ]
     needs: [ tag, compute-node-image ]
     strategy:
       fail-fast: false
@@ -650,31 +596,34 @@ jobs:
       run:
         shell: sh -eu {0}
     env:
-      VM_BUILDER_VERSION: v0.4.6
+      VM_INFORMANT_VERSION: 0.1.1
 
     steps:
-      - name: Checkout
-        uses: actions/checkout@v1
-        with:
-          fetch-depth: 0
-
-      - name: Downloading vm-builder
+      - name: Downloading latest vm-builder
         run: |
-          curl -L https://github.com/neondatabase/neonvm/releases/download/$VM_BUILDER_VERSION/vm-builder -o vm-builder
+          curl -L https://github.com/neondatabase/neonvm/releases/latest/download/vm-builder -o vm-builder
           chmod +x vm-builder
 
       - name: Pulling compute-node image
         run: |
           docker pull 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-${{ matrix.version }}:${{needs.tag.outputs.build-tag}}
 
-      - name: Building VM compute-node rootfs
+      - name: Downloading VM informant version ${{ env.VM_INFORMANT_VERSION }}
         run: |
-          docker build -t temp-vm-compute-node --build-arg SRC_IMAGE=369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-${{ matrix.version }}:${{needs.tag.outputs.build-tag}} -f Dockerfile.vm-compute-node .
+          curl -fL https://github.com/neondatabase/autoscaling/releases/download/${{ env.VM_INFORMANT_VERSION }}/vm-informant -o vm-informant
+          chmod +x vm-informant
+
+      - name: Adding VM informant to compute-node image
+        run: |
+          ID=$(docker create 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-${{ matrix.version }}:${{needs.tag.outputs.build-tag}})
+          docker cp vm-informant $ID:/bin/vm-informant
+          docker commit $ID temp-vm-compute-node
+          docker rm -f $ID
 
       - name: Build vm image
         run: |
           # note: as of 2023-01-12, vm-builder requires a trailing ":latest" for local images
-          ./vm-builder -use-inittab -src=temp-vm-compute-node:latest -dst=369495373322.dkr.ecr.eu-central-1.amazonaws.com/vm-compute-node-${{ matrix.version }}:${{needs.tag.outputs.build-tag}}
+          ./vm-builder -src=temp-vm-compute-node:latest -dst=369495373322.dkr.ecr.eu-central-1.amazonaws.com/vm-compute-node-${{ matrix.version }}:${{needs.tag.outputs.build-tag}}
 
       - name: Pushing vm-compute-node image
         run: |
@@ -682,7 +631,7 @@ jobs:
 
   test-images:
     needs: [ tag, neon-image, compute-node-image, compute-tools-image ]
-    runs-on: [ self-hosted, gen3, small ]
+    runs-on: [ self-hosted, dev, x64 ]
 
     steps:
       - name: Checkout
@@ -724,39 +673,20 @@ jobs:
           docker compose -f ./docker-compose/docker-compose.yml down
 
   promote-images:
-    runs-on: [ self-hosted, gen3, small ]
+    runs-on: [ self-hosted, dev, x64 ]
     needs: [ tag, test-images, vm-compute-node-image ]
-    container: golang:1.19-bullseye
     if: github.event_name != 'workflow_dispatch'
+    container: amazon/aws-cli
+    strategy:
+      fail-fast: false
+      matrix:
+        name: [ neon, compute-node-v14, vm-compute-node-v14, compute-node-v15, vm-compute-node-v15, compute-tools]
 
     steps:
-      - name: Install Crane & ECR helper
-        if: |
-          (github.ref_name == 'main' || github.ref_name == 'release') &&
-          github.event_name != 'workflow_dispatch'
+      - name: Promote image to latest
         run: |
-          go install github.com/google/go-containerregistry/cmd/crane@31786c6cbb82d6ec4fb8eb79cd9387905130534e # v0.11.0
-          go install github.com/awslabs/amazon-ecr-credential-helper/ecr-login/cli/docker-credential-ecr-login@69c85dc22db6511932bbf119e1a0cc5c90c69a7f # v0.6.0
-
-      - name: Configure ECR login
-        run: |
-          mkdir /github/home/.docker/
-          echo "{\"credsStore\":\"ecr-login\"}" > /github/home/.docker/config.json
-
-      - name: Add latest tag to images
-        if: |
-          (github.ref_name == 'main' || github.ref_name == 'release') &&
-          github.event_name != 'workflow_dispatch'
-        run: |
-          crane tag 369495373322.dkr.ecr.eu-central-1.amazonaws.com/neon:${{needs.tag.outputs.build-tag}} latest
-          crane tag 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-tools:${{needs.tag.outputs.build-tag}} latest
-          crane tag 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v14:${{needs.tag.outputs.build-tag}} latest
-          crane tag 369495373322.dkr.ecr.eu-central-1.amazonaws.com/vm-compute-node-v14:${{needs.tag.outputs.build-tag}} latest
-          crane tag 369495373322.dkr.ecr.eu-central-1.amazonaws.com/compute-node-v15:${{needs.tag.outputs.build-tag}} latest
-          crane tag 369495373322.dkr.ecr.eu-central-1.amazonaws.com/vm-compute-node-v15:${{needs.tag.outputs.build-tag}} latest
-
-      - name: Cleanup ECR folder
-        run: rm -rf ~/.ecr
+          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: [ self-hosted, dev, x64 ]
@@ -846,11 +776,114 @@ jobs:
           crane tag neondatabase/compute-node-v15:${{needs.tag.outputs.build-tag}} latest
           crane tag neondatabase/vm-compute-node-v15:${{needs.tag.outputs.build-tag}} latest
 
-      - name: Cleanup ECR folder
-        run: rm -rf ~/.ecr
+  calculate-deploy-targets:
+    runs-on: [ self-hosted, dev, x64 ]
+    if: |
+      github.ref_name == 'release' &&
+      github.event_name != 'workflow_dispatch'
+    outputs:
+      matrix-include: ${{ steps.set-matrix.outputs.include }}
+    steps:
+      - id: set-matrix
+        run: |
+          if [[ "$GITHUB_REF_NAME" == "release" ]]; then
+            PRODUCTION='{"env_name": "production", "proxy_job": "neon-proxy", "proxy_config": "production.proxy", "storage_broker_ns": "neon-storage-broker", "storage_broker_config": "production.neon-storage-broker", "kubeconfig_secret": "PRODUCTION_KUBECONFIG_DATA", "console_api_key_secret": "NEON_PRODUCTION_API_KEY"}'
+            echo "include=[$PRODUCTION]" >> $GITHUB_OUTPUT
+          else
+            echo "GITHUB_REF_NAME (value '$GITHUB_REF_NAME') is not set to 'release'"
+            exit 1
+          fi
+
+  deploy:
+    runs-on: [ self-hosted, dev, x64 ]
+    container: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/ansible:pinned
+    # We need both storage **and** compute images for deploy, because control plane picks the compute version based on the storage version.
+    # If it notices a fresh storage it may bump the compute version. And if compute image failed to build it may break things badly
+    needs: [ push-docker-hub, calculate-deploy-targets, tag, regress-tests ]
+    if: |
+      github.ref_name == 'release' &&
+      github.event_name != 'workflow_dispatch'
+    defaults:
+      run:
+        shell: bash
+    strategy:
+      matrix:
+        include: ${{fromJSON(needs.calculate-deploy-targets.outputs.matrix-include)}}
+    environment:
+      name: prod-old
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v3
+        with:
+          submodules: true
+          fetch-depth: 0
+
+      - name: Redeploy
+        run: |
+          export DOCKER_TAG=${{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
+
+          eval $(ssh-agent)
+          echo "${{ secrets.TELEPORT_SSH_KEY }}"  | tr -d '\n'| base64 --decode >ssh-key
+          echo "${{ secrets.TELEPORT_SSH_CERT }}" | tr -d '\n'| base64 --decode >ssh-key-cert.pub
+          chmod 0600 ssh-key
+          ssh-add ssh-key
+          rm -f ssh-key ssh-key-cert.pub
+          ANSIBLE_CONFIG=./ansible.cfg ansible-galaxy collection install sivel.toiletwater
+          ANSIBLE_CONFIG=./ansible.cfg ansible-playbook deploy.yaml -i ${{ matrix.env_name }}.hosts.yaml -e CONSOLE_API_TOKEN=${{ secrets[matrix.console_api_key_secret] }} -e SENTRY_URL_PAGESERVER=${{ secrets.SENTRY_URL_PAGESERVER }} -e SENTRY_URL_SAFEKEEPER=${{ secrets.SENTRY_URL_SAFEKEEPER }}
+          rm -f neon_install.tar.gz .neon_current_version
+
+  deploy-new:
+    runs-on: [ self-hosted, dev, x64 ]
+    container: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/ansible:pinned
+    # We need both storage **and** compute images for deploy, because control plane picks the compute version based on the storage version.
+    # If it notices a fresh storage it may bump the compute version. And if compute image failed to build it may break things badly
+    needs: [ push-docker-hub, tag, regress-tests ]
+    if: |
+      (github.ref_name == 'main') &&
+      github.event_name != 'workflow_dispatch'
+    defaults:
+      run:
+        shell: bash
+    strategy:
+      matrix:
+        target_region: [ eu-west-1, us-east-2 ]
+    environment:
+      name: dev-${{ matrix.target_region }}
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v3
+        with:
+          submodules: true
+          fetch-depth: 0
+
+      - name: Redeploy
+        run: |
+          export DOCKER_TAG=${{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 }} -e SENTRY_URL_PAGESERVER=${{ secrets.SENTRY_URL_PAGESERVER }} -e SENTRY_URL_SAFEKEEPER=${{ secrets.SENTRY_URL_SAFEKEEPER }}
+          rm -f neon_install.tar.gz .neon_current_version
 
   deploy-pr-test-new:
-    runs-on: [ self-hosted, gen3, small ]
+    runs-on: [ self-hosted, dev, x64 ]
     container: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/ansible:pinned
     # We need both storage **and** compute images for deploy, because control plane picks the compute version based on the storage version.
     # If it notices a fresh storage it may bump the compute version. And if compute image failed to build it may break things badly
@@ -882,40 +915,311 @@ jobs:
           ansible-playbook deploy.yaml -i staging.${{ matrix.target_region }}.hosts.yaml -e @ssm_config -e CONSOLE_API_TOKEN=${{ secrets.NEON_STAGING_API_KEY }} -e SENTRY_URL_PAGESERVER=${{ secrets.SENTRY_URL_PAGESERVER }} -e SENTRY_URL_SAFEKEEPER=${{ secrets.SENTRY_URL_SAFEKEEPER }}
           rm -f neon_install.tar.gz .neon_current_version
 
-      - name: Cleanup ansible folder
-        run: rm -rf ~/.ansible
-
-  deploy:
-    runs-on: [ self-hosted, gen3, small ]
-    container: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/ansible:latest
+  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, tag, regress-tests ]
-    if: ( github.ref_name == 'main' || github.ref_name == 'release' ) && github.event_name != 'workflow_dispatch'
+    if: |
+      (github.ref_name == 'release') &&
+      github.event_name != 'workflow_dispatch'
+    defaults:
+      run:
+        shell: bash
+    strategy:
+      matrix:
+        target_region: [ us-east-2, us-west-2, eu-central-1, ap-southeast-1 ]
+    environment:
+      name: prod-${{ matrix.target_region }}
     steps:
       - name: Checkout
         uses: actions/checkout@v3
         with:
-          submodules: false
+          submodules: true
           fetch-depth: 0
 
-      - name: Trigger deploy workflow
-        env:
-          GH_TOKEN: ${{ github.token }}
+      - name: Redeploy
         run: |
+          export DOCKER_TAG=${{needs.tag.outputs.build-tag}}
+          cd "$(pwd)/.github/ansible"
+
           if [[ "$GITHUB_REF_NAME" == "main" ]]; then
-            gh workflow run deploy-dev.yml --ref main -f branch=${{ github.sha }} -f dockerTag=${{needs.tag.outputs.build-tag}}
+            ./get_binaries.sh
           elif [[ "$GITHUB_REF_NAME" == "release" ]]; then
-            gh workflow run deploy-prod.yml --ref release -f branch=${{ github.sha }} -f dockerTag=${{needs.tag.outputs.build-tag}} -f disclamerAcknowledged=true
+            RELEASE=true ./get_binaries.sh
           else
             echo "GITHUB_REF_NAME (value '$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 }} -e SENTRY_URL_PAGESERVER=${{ secrets.SENTRY_URL_PAGESERVER }} -e SENTRY_URL_SAFEKEEPER=${{ secrets.SENTRY_URL_SAFEKEEPER }}
+          rm -f neon_install.tar.gz .neon_current_version
+
+  deploy-proxy:
+    runs-on: [ self-hosted, dev, x64 ]
+    container: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/base: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 == 'release' &&
+      github.event_name != 'workflow_dispatch'
+    defaults:
+      run:
+        shell: bash
+    strategy:
+      matrix:
+        include: ${{fromJSON(needs.calculate-deploy-targets.outputs.matrix-include)}}
+    environment:
+      name: prod-old
+    env:
+      KUBECONFIG: .kubeconfig
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v3
+        with:
+          submodules: true
+          fetch-depth: 0
+
+      - name: Add curl
+        run: apt update && apt install curl -y
+
+      - name: Store kubeconfig file
+        run: |
+          echo "${{ secrets[matrix.kubeconfig_secret] }}" | base64 --decode > ${KUBECONFIG}
+          chmod 0600 ${KUBECONFIG}
+
+      - name: Setup helm v3
+        run: |
+          curl -s https://raw.githubusercontent.com/helm/helm/main/scripts/get-helm-3 | bash
+          helm repo add neondatabase https://neondatabase.github.io/helm-charts
+
+      - name: Re-deploy proxy
+        run: |
+          DOCKER_TAG=${{needs.tag.outputs.build-tag}}
+          helm upgrade ${{ matrix.proxy_job }}       neondatabase/neon-proxy --namespace neon-proxy --install --atomic -f .github/helm-values/${{ matrix.proxy_config }}.yaml       --set image.tag=${DOCKER_TAG} --set settings.sentryUrl=${{ secrets.SENTRY_URL_PROXY }} --wait --timeout 15m0s
+          helm upgrade ${{ matrix.proxy_job }}-scram neondatabase/neon-proxy --namespace neon-proxy --install --atomic -f .github/helm-values/${{ matrix.proxy_config }}-scram.yaml --set image.tag=${DOCKER_TAG} --set settings.sentryUrl=${{ secrets.SENTRY_URL_PROXY }} --wait --timeout 15m0s
+
+  deploy-storage-broker:
+    name: deploy storage broker on old staging and old prod
+    runs-on: [ self-hosted, dev, x64 ]
+    container: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/base: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 == 'release' &&
+      github.event_name != 'workflow_dispatch'
+    defaults:
+      run:
+        shell: bash
+    strategy:
+      matrix:
+        include: ${{fromJSON(needs.calculate-deploy-targets.outputs.matrix-include)}}
+    environment:
+      name: prod-old
+    env:
+      KUBECONFIG: .kubeconfig
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v3
+        with:
+          submodules: true
+          fetch-depth: 0
+
+      - name: Add curl
+        run: apt update && apt install curl -y
+
+      - name: Store kubeconfig file
+        run: |
+          echo "${{ secrets[matrix.kubeconfig_secret] }}" | base64 --decode > ${KUBECONFIG}
+          chmod 0600 ${KUBECONFIG}
+
+      - name: Setup helm v3
+        run: |
+          curl -s https://raw.githubusercontent.com/helm/helm/main/scripts/get-helm-3 | bash
+          helm repo add neondatabase https://neondatabase.github.io/helm-charts
+
+      - name: Deploy storage-broker
+        run:
+          helm upgrade neon-storage-broker neondatabase/neon-storage-broker --namespace ${{ matrix.storage_broker_ns }} --create-namespace --install --atomic -f .github/helm-values/${{ matrix.storage_broker_config }}.yaml --set image.tag=${{ needs.tag.outputs.build-tag }} --set settings.sentryUrl=${{ secrets.SENTRY_URL_BROKER }} --wait --timeout 5m0s
+
+  deploy-proxy-new:
+    runs-on: [ self-hosted, dev, x64 ]
+    container: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/ansible:pinned
+    # Compute image isn't strictly required for proxy deploy, but let's still wait for it to run all deploy jobs consistently.
+    needs: [ push-docker-hub, tag, regress-tests ]
+    if: |
+      (github.ref_name == 'main') &&
+      github.event_name != 'workflow_dispatch'
+    defaults:
+      run:
+        shell: bash
+    strategy:
+      matrix:
+        include:
+          - target_region:  us-east-2
+            target_cluster: dev-us-east-2-beta
+            deploy_link_proxy: true
+            deploy_legacy_scram_proxy: true
+          - target_region:  eu-west-1
+            target_cluster: dev-eu-west-1-zeta
+            deploy_link_proxy: false
+            deploy_legacy_scram_proxy: false
+    environment:
+      name: dev-${{ matrix.target_region }}
+    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  ${{ matrix.target_cluster }}
+
+      - name: Re-deploy scram proxy
+        run: |
+          DOCKER_TAG=${{needs.tag.outputs.build-tag}}
+          helm upgrade neon-proxy-scram neondatabase/neon-proxy --namespace neon-proxy --create-namespace --install --atomic -f .github/helm-values/${{ matrix.target_cluster }}.neon-proxy-scram.yaml --set image.tag=${DOCKER_TAG} --set settings.sentryUrl=${{ secrets.SENTRY_URL_PROXY }} --wait --timeout 15m0s
+
+      - name: Re-deploy link proxy
+        if: matrix.deploy_link_proxy
+        run: |
+          DOCKER_TAG=${{needs.tag.outputs.build-tag}}
+          helm upgrade neon-proxy-link neondatabase/neon-proxy --namespace neon-proxy --create-namespace --install --atomic -f .github/helm-values/${{ matrix.target_cluster }}.neon-proxy-link.yaml --set image.tag=${DOCKER_TAG} --set settings.sentryUrl=${{ secrets.SENTRY_URL_PROXY }} --wait --timeout 15m0s
+
+      - name: Re-deploy legacy scram proxy
+        if: matrix.deploy_legacy_scram_proxy
+        run: |
+          DOCKER_TAG=${{needs.tag.outputs.build-tag}}
+          helm upgrade neon-proxy-scram-legacy neondatabase/neon-proxy --namespace neon-proxy --create-namespace --install --atomic -f .github/helm-values/${{ matrix.target_cluster }}.neon-proxy-scram-legacy.yaml --set image.tag=${DOCKER_TAG} --set settings.sentryUrl=${{ secrets.SENTRY_URL_PROXY }} --wait --timeout 15m0s
+
+  deploy-storage-broker-dev-new:
+    runs-on: [ self-hosted, dev, x64 ]
+    container: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/ansible:pinned
+    # Compute image isn't strictly required for proxy deploy, but let's still wait for it to run all deploy jobs consistently.
+    needs: [ push-docker-hub, tag, regress-tests ]
+    if: |
+      (github.ref_name == 'main') &&
+      github.event_name != 'workflow_dispatch'
+    defaults:
+      run:
+        shell: bash
+    strategy:
+      matrix:
+        include:
+          - target_region:  us-east-2
+            target_cluster: dev-us-east-2-beta
+          - target_region:  eu-west-1
+            target_cluster: dev-eu-west-1-zeta
+    environment:
+      name: dev-${{ matrix.target_region }}
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v3
+        with:
+          submodules: true
+          fetch-depth: 0
+
+      - name: Configure environment
+        run: |
+          helm repo add neondatabase https://neondatabase.github.io/helm-charts
+          aws --region ${{ matrix.target_region }} eks update-kubeconfig --name  ${{ matrix.target_cluster }}
+
+      - name: Deploy storage-broker
+        run:
+          helm upgrade neon-storage-broker-lb neondatabase/neon-storage-broker --namespace neon-storage-broker-lb --create-namespace --install --atomic -f .github/helm-values/${{ matrix.target_cluster }}.neon-storage-broker.yaml --set image.tag=${{ needs.tag.outputs.build-tag }} --set settings.sentryUrl=${{ secrets.SENTRY_URL_BROKER }} --wait --timeout 5m0s
+
+  deploy-proxy-prod-new:
+    runs-on: prod
+    container: 093970136003.dkr.ecr.eu-central-1.amazonaws.com/ansible:latest
+    # Compute image isn't strictly required for proxy deploy, but let's still wait for it to run all deploy jobs consistently.
+    needs: [ push-docker-hub, tag, regress-tests ]
+    if: |
+      (github.ref_name == 'release') &&
+      github.event_name != 'workflow_dispatch'
+    defaults:
+      run:
+        shell: bash
+    strategy:
+      matrix:
+        include:
+          - target_region:  us-east-2
+            target_cluster: prod-us-east-2-delta
+          - target_region:  us-west-2
+            target_cluster: prod-us-west-2-eta
+          - target_region: eu-central-1
+            target_cluster: prod-eu-central-1-gamma
+          - target_region: ap-southeast-1
+            target_cluster: prod-ap-southeast-1-epsilon
+    environment:
+      name: prod-${{ matrix.target_region }}
+    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 --atomic -f .github/helm-values/${{ matrix.target_cluster }}.neon-proxy-scram.yaml --set image.tag=${DOCKER_TAG} --set settings.sentryUrl=${{ secrets.SENTRY_URL_PROXY }} --wait --timeout 15m0s
+
+  deploy-storage-broker-prod-new:
+    runs-on: prod
+    container: 093970136003.dkr.ecr.eu-central-1.amazonaws.com/ansible:latest
+    # Compute image isn't strictly required for proxy deploy, but let's still wait for it to run all deploy jobs consistently.
+    needs: [ push-docker-hub, tag, regress-tests ]
+    if: |
+      (github.ref_name == 'release') &&
+      github.event_name != 'workflow_dispatch'
+    defaults:
+      run:
+        shell: bash
+    strategy:
+      matrix:
+        include:
+          - target_region:  us-east-2
+            target_cluster: prod-us-east-2-delta
+          - target_region:  us-west-2
+            target_cluster: prod-us-west-2-eta
+          - target_region: eu-central-1
+            target_cluster: prod-eu-central-1-gamma
+          - target_region: ap-southeast-1
+            target_cluster: prod-ap-southeast-1-epsilon
+    environment:
+      name: prod-${{ matrix.target_region }}
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v3
+        with:
+          submodules: true
+          fetch-depth: 0
+
+      - name: Configure environment
+        run: |
+          helm repo add neondatabase https://neondatabase.github.io/helm-charts
+          aws --region ${{ matrix.target_region }} eks update-kubeconfig --name  ${{ matrix.target_cluster }}
+
+      - name: Deploy storage-broker
+        run:
+          helm upgrade neon-storage-broker-lb neondatabase/neon-storage-broker --namespace neon-storage-broker-lb --create-namespace --install --atomic -f .github/helm-values/${{ matrix.target_cluster }}.neon-storage-broker.yaml --set image.tag=${{ needs.tag.outputs.build-tag }} --set settings.sentryUrl=${{ secrets.SENTRY_URL_BROKER }} --wait --timeout 5m0s
+
   promote-compatibility-data:
-    runs-on: [ self-hosted, gen3, small ]
+    runs-on: [ self-hosted, dev, x64 ]
     container:
       image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned
       options: --init
-    needs: [ push-docker-hub, tag, regress-tests ]
+    needs: [ deploy, deploy-proxy ]
     if: github.ref_name == 'release' && github.event_name != 'workflow_dispatch'
     steps:
       - name: Promote compatibility snapshot for the release

.github/workflows/deploy-dev.yml

@@ -1,179 +0,0 @@
-name: Neon Deploy dev
-
-on:
-  workflow_dispatch:
-    inputs:
-      dockerTag:
-        description: 'Docker tag to deploy'
-        required: true
-        type: string
-      branch:
-        description: 'Branch or commit used for deploy scripts and configs'
-        required: true
-        type: string
-        default: 'main'
-      deployStorage:
-        description: 'Deploy storage'
-        required: true
-        type: boolean
-        default: true
-      deployProxy:
-        description: 'Deploy proxy'
-        required: true
-        type: boolean
-        default: true
-      deployStorageBroker:
-        description: 'Deploy storage-broker'
-        required: true
-        type: boolean
-        default: true
-
-env:
-  AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_DEV }}
-  AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_KEY_DEV }}
-
-concurrency:
-  group: deploy-dev
-  cancel-in-progress: false
-
-jobs:
-  deploy-storage-new:
-    runs-on: [ self-hosted, gen3, small ]
-    container:
-      image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/ansible:pinned
-      options: --user root --privileged
-    if: inputs.deployStorage
-    defaults:
-      run:
-        shell: bash
-    strategy:
-      matrix:
-        target_region: [ eu-west-1, us-east-2 ]
-    environment:
-      name: dev-${{ matrix.target_region }}
-    steps:
-      - name: Checkout
-        uses: actions/checkout@v3
-        with:
-          submodules: true
-          fetch-depth: 0
-          ref: ${{ inputs.branch }}
-
-      - name: Redeploy
-        run: |
-          export DOCKER_TAG=${{ inputs.dockerTag }}
-          cd "$(pwd)/.github/ansible"
-
-          ./get_binaries.sh
-
-          ansible-galaxy collection install sivel.toiletwater
-          ansible-playbook -v deploy.yaml -i staging.${{ matrix.target_region }}.hosts.yaml -e @ssm_config -e CONSOLE_API_TOKEN=${{ secrets.NEON_STAGING_API_KEY }} -e SENTRY_URL_PAGESERVER=${{ secrets.SENTRY_URL_PAGESERVER }} -e SENTRY_URL_SAFEKEEPER=${{ secrets.SENTRY_URL_SAFEKEEPER }}
-          rm -f neon_install.tar.gz .neon_current_version
-
-      - name: Cleanup ansible folder
-        run: rm -rf ~/.ansible
-
-  deploy-proxy-new:
-    runs-on: [ self-hosted, gen3, small ]
-    container: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/ansible:pinned
-    if: inputs.deployProxy
-    defaults:
-      run:
-        shell: bash
-    strategy:
-      matrix:
-        include:
-          - target_region:  us-east-2
-            target_cluster: dev-us-east-2-beta
-            deploy_link_proxy: true
-            deploy_legacy_scram_proxy: true
-          - target_region:  eu-west-1
-            target_cluster: dev-eu-west-1-zeta
-            deploy_link_proxy: false
-            deploy_legacy_scram_proxy: false
-    environment:
-      name: dev-${{ matrix.target_region }}
-    steps:
-      - name: Checkout
-        uses: actions/checkout@v3
-        with:
-          submodules: true
-          fetch-depth: 0
-          ref: ${{ inputs.branch }}
-  
-      - name: Configure AWS Credentials
-        uses: aws-actions/configure-aws-credentials@v1-node16
-        with:
-          role-to-assume: arn:aws:iam::369495373322:role/github-runner
-          aws-region: eu-central-1
-          role-skip-session-tagging: true
-          role-duration-seconds: 1800
-  
-      - name: Configure environment
-        run: |
-          helm repo add neondatabase https://neondatabase.github.io/helm-charts
-          aws --region ${{ matrix.target_region }} eks update-kubeconfig --name  ${{ matrix.target_cluster }}
-  
-      - name: Re-deploy scram proxy
-        run: |
-          DOCKER_TAG=${{ inputs.dockerTag }}
-          helm upgrade neon-proxy-scram neondatabase/neon-proxy --namespace neon-proxy --create-namespace --install --atomic -f .github/helm-values/${{ matrix.target_cluster }}.neon-proxy-scram.yaml --set image.tag=${DOCKER_TAG} --set settings.sentryUrl=${{ secrets.SENTRY_URL_PROXY }} --wait --timeout 15m0s
-  
-      - name: Re-deploy link proxy
-        if: matrix.deploy_link_proxy
-        run: |
-          DOCKER_TAG=${{ inputs.dockerTag }}
-          helm upgrade neon-proxy-link neondatabase/neon-proxy --namespace neon-proxy --create-namespace --install --atomic -f .github/helm-values/${{ matrix.target_cluster }}.neon-proxy-link.yaml --set image.tag=${DOCKER_TAG} --set settings.sentryUrl=${{ secrets.SENTRY_URL_PROXY }} --wait --timeout 15m0s
-  
-      - name: Re-deploy legacy scram proxy
-        if: matrix.deploy_legacy_scram_proxy
-        run: |
-          DOCKER_TAG=${{ inputs.dockerTag }}
-          helm upgrade neon-proxy-scram-legacy neondatabase/neon-proxy --namespace neon-proxy --create-namespace --install --atomic -f .github/helm-values/${{ matrix.target_cluster }}.neon-proxy-scram-legacy.yaml --set image.tag=${DOCKER_TAG} --set settings.sentryUrl=${{ secrets.SENTRY_URL_PROXY }} --wait --timeout 15m0s
-  
-      - name: Cleanup helm folder
-        run: rm -rf ~/.cache
-  
-  deploy-storage-broker-new:
-    runs-on: [ self-hosted, gen3, small ]
-    container: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/ansible:pinned
-    if: inputs.deployStorageBroker
-    defaults:
-      run:
-        shell: bash
-    strategy:
-      matrix:
-        include:
-          - target_region:  us-east-2
-            target_cluster: dev-us-east-2-beta
-          - target_region:  eu-west-1
-            target_cluster: dev-eu-west-1-zeta
-    environment:
-      name: dev-${{ matrix.target_region }}
-    steps:
-      - name: Checkout
-        uses: actions/checkout@v3
-        with:
-          submodules: true
-          fetch-depth: 0
-          ref: ${{ inputs.branch }}
-  
-      - name: Configure AWS Credentials
-        uses: aws-actions/configure-aws-credentials@v1-node16
-        with:
-          role-to-assume: arn:aws:iam::369495373322:role/github-runner
-          aws-region: eu-central-1
-          role-skip-session-tagging: true
-          role-duration-seconds: 1800
-  
-      - name: Configure environment
-        run: |
-          helm repo add neondatabase https://neondatabase.github.io/helm-charts
-          aws --region ${{ matrix.target_region }} eks update-kubeconfig --name  ${{ matrix.target_cluster }}
-  
-      - name: Deploy storage-broker
-        run:
-          helm upgrade neon-storage-broker-lb neondatabase/neon-storage-broker --namespace neon-storage-broker-lb --create-namespace --install --atomic -f .github/helm-values/${{ matrix.target_cluster }}.neon-storage-broker.yaml --set image.tag=${{ inputs.dockerTag }} --set settings.sentryUrl=${{ secrets.SENTRY_URL_BROKER }} --wait --timeout 5m0s
-  
-      - name: Cleanup helm folder
-        run: rm -rf ~/.cache

.github/workflows/deploy-prod.yml

@@ -1,167 +0,0 @@
-name: Neon Deploy prod
-
-on:
-  workflow_dispatch:
-    inputs:
-      dockerTag:
-        description: 'Docker tag to deploy'
-        required: true
-        type: string
-      branch:
-        description: 'Branch or commit used for deploy scripts and configs'
-        required: true
-        type: string
-        default: 'release'
-      deployStorage:
-        description: 'Deploy storage'
-        required: true
-        type: boolean
-        default: true
-      deployProxy:
-        description: 'Deploy proxy'
-        required: true
-        type: boolean
-        default: true
-      deployStorageBroker:
-        description: 'Deploy storage-broker'
-        required: true
-        type: boolean
-        default: true
-      disclamerAcknowledged:
-        description: 'I confirm that there is an emergency and I can not use regular release workflow'
-        required: true
-        type: boolean
-        default: false
-
-concurrency:
-  group: deploy-prod
-  cancel-in-progress: false
-
-jobs:
-  deploy-prod-new:
-    runs-on: prod
-    container:
-      image: 093970136003.dkr.ecr.eu-central-1.amazonaws.com/ansible:latest
-      options: --user root --privileged
-    if: inputs.deployStorage && inputs.disclamerAcknowledged
-    defaults:
-      run:
-        shell: bash
-    strategy:
-      matrix:
-        target_region: [ us-east-2, us-west-2, eu-central-1, ap-southeast-1 ]
-    environment:
-      name: prod-${{ matrix.target_region }}
-    steps:
-      - name: Checkout
-        uses: actions/checkout@v3
-        with:
-          submodules: true
-          fetch-depth: 0
-          ref: ${{ inputs.branch }}
-
-      - name: Redeploy
-        run: |
-          export DOCKER_TAG=${{ inputs.dockerTag }}
-          cd "$(pwd)/.github/ansible"
-
-          ./get_binaries.sh
-
-          ansible-galaxy collection install sivel.toiletwater
-          ansible-playbook -v deploy.yaml -i prod.${{ matrix.target_region }}.hosts.yaml -e @ssm_config -e CONSOLE_API_TOKEN=${{ secrets.NEON_PRODUCTION_API_KEY }} -e SENTRY_URL_PAGESERVER=${{ secrets.SENTRY_URL_PAGESERVER }} -e SENTRY_URL_SAFEKEEPER=${{ secrets.SENTRY_URL_SAFEKEEPER }}
-          rm -f neon_install.tar.gz .neon_current_version
-
-  deploy-proxy-prod-new:
-    runs-on: prod
-    container: 093970136003.dkr.ecr.eu-central-1.amazonaws.com/ansible:latest
-    if: inputs.deployProxy && inputs.disclamerAcknowledged
-    defaults:
-      run:
-        shell: bash
-    strategy:
-      matrix:
-        include:
-          - target_region:  us-east-2
-            target_cluster: prod-us-east-2-delta
-            deploy_link_proxy: true
-            deploy_legacy_scram_proxy: false
-          - target_region:  us-west-2
-            target_cluster: prod-us-west-2-eta
-            deploy_link_proxy: false
-            deploy_legacy_scram_proxy: true
-          - target_region: eu-central-1
-            target_cluster: prod-eu-central-1-gamma
-            deploy_link_proxy: false
-            deploy_legacy_scram_proxy: false
-          - target_region: ap-southeast-1
-            target_cluster: prod-ap-southeast-1-epsilon
-            deploy_link_proxy: false
-            deploy_legacy_scram_proxy: false
-    environment:
-      name: prod-${{ matrix.target_region }}
-    steps:
-      - name: Checkout
-        uses: actions/checkout@v3
-        with:
-          submodules: true
-          fetch-depth: 0
-          ref: ${{ inputs.branch }}
-
-      - name: Configure environment
-        run: |
-          helm repo add neondatabase https://neondatabase.github.io/helm-charts
-          aws --region ${{ matrix.target_region }} eks update-kubeconfig --name  ${{ matrix.target_cluster }}
-
-      - name: Re-deploy scram proxy
-        run: |
-          DOCKER_TAG=${{ inputs.dockerTag }}
-          helm upgrade neon-proxy-scram neondatabase/neon-proxy --namespace neon-proxy --create-namespace --install --atomic -f .github/helm-values/${{ matrix.target_cluster }}.neon-proxy-scram.yaml --set image.tag=${DOCKER_TAG} --set settings.sentryUrl=${{ secrets.SENTRY_URL_PROXY }} --wait --timeout 15m0s
-
-      - name: Re-deploy link proxy
-        if: matrix.deploy_link_proxy
-        run: |
-          DOCKER_TAG=${{ inputs.dockerTag }}
-          helm upgrade neon-proxy-link neondatabase/neon-proxy --namespace neon-proxy --create-namespace --install --atomic -f .github/helm-values/${{ matrix.target_cluster }}.neon-proxy-link.yaml --set image.tag=${DOCKER_TAG} --set settings.sentryUrl=${{ secrets.SENTRY_URL_PROXY }} --wait --timeout 15m0s
-
-      - name: Re-deploy legacy scram proxy
-        if: matrix.deploy_legacy_scram_proxy
-        run: |
-          DOCKER_TAG=${{ inputs.dockerTag }}
-          helm upgrade neon-proxy-scram-legacy neondatabase/neon-proxy --namespace neon-proxy --create-namespace --install --atomic -f .github/helm-values/${{ matrix.target_cluster }}.neon-proxy-scram-legacy.yaml --set image.tag=${DOCKER_TAG} --set settings.sentryUrl=${{ secrets.SENTRY_URL_PROXY }} --wait --timeout 15m0s
-
-  deploy-storage-broker-prod-new:
-    runs-on: prod
-    container: 093970136003.dkr.ecr.eu-central-1.amazonaws.com/ansible:latest
-    if: inputs.deployStorageBroker && inputs.disclamerAcknowledged
-    defaults:
-      run:
-        shell: bash
-    strategy:
-      matrix:
-        include:
-          - target_region:  us-east-2
-            target_cluster: prod-us-east-2-delta
-          - target_region:  us-west-2
-            target_cluster: prod-us-west-2-eta
-          - target_region: eu-central-1
-            target_cluster: prod-eu-central-1-gamma
-          - target_region: ap-southeast-1
-            target_cluster: prod-ap-southeast-1-epsilon
-    environment:
-      name: prod-${{ matrix.target_region }}
-    steps:
-      - name: Checkout
-        uses: actions/checkout@v3
-        with:
-          submodules: true
-          fetch-depth: 0
-          ref: ${{ inputs.branch }}
-
-      - name: Configure environment
-        run: |
-          helm repo add neondatabase https://neondatabase.github.io/helm-charts
-          aws --region ${{ matrix.target_region }} eks update-kubeconfig --name  ${{ matrix.target_cluster }}
-
-      - name: Deploy storage-broker
-        run:
-          helm upgrade neon-storage-broker-lb neondatabase/neon-storage-broker --namespace neon-storage-broker-lb --create-namespace --install --atomic -f .github/helm-values/${{ matrix.target_cluster }}.neon-storage-broker.yaml --set image.tag=${{ inputs.dockerTag }} --set settings.sentryUrl=${{ secrets.SENTRY_URL_BROKER }} --wait --timeout 5m0s

.github/workflows/neon_extra_builds.yml

@@ -4,7 +4,6 @@ on:
   push:
     branches:
     - main
-  pull_request:
 
 defaults:
   run:
@@ -21,7 +20,6 @@ env:
 
 jobs:
   check-macos-build:
-    if: github.ref_name == 'main' || contains(github.event.pull_request.labels.*.name, 'run-extra-build-macos')
     timeout-minutes: 90
     runs-on: macos-latest
 
@@ -95,16 +93,11 @@ jobs:
         run: ./run_clippy.sh
 
   gather-rust-build-stats:
-    if: github.ref_name == 'main' || contains(github.event.pull_request.labels.*.name, 'run-extra-build-stats')
-    runs-on: [ self-hosted, gen3, large ]
-    container:
-      image: 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:pinned
-      options: --init
+    timeout-minutes: 90
+    runs-on: ubuntu-latest
 
     env:
       BUILD_TYPE: release
-      # remove the cachepot wrapper and build without crate caches
-      RUSTC_WRAPPER: ""
       # build with incremental compilation produce partial results
       # so do not attempt to cache this build, also disable the incremental compilation
       CARGO_INCREMENTAL: 0
@@ -116,6 +109,11 @@ jobs:
           submodules: true
           fetch-depth: 1
 
+      - name: Install Ubuntu postgres dependencies
+        run: |
+          sudo apt update
+          sudo apt install build-essential libreadline-dev zlib1g-dev flex bison libseccomp-dev libssl-dev protobuf-compiler
+
       # Some of our rust modules use FFI and need those to be checked
       - name: Get postgres headers
         run: make postgres-headers -j$(nproc)
@@ -124,31 +122,7 @@ jobs:
         run: cargo build --all --release --timings
 
       - name: Upload the build stats
-        id: upload-stats
-        env:
-          BUCKET: neon-github-public-dev
-          SHA: ${{ github.event.pull_request.head.sha || github.sha }}
-          AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_DEV }}
-          AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_KEY_DEV }}
-        run: |
-          REPORT_URL=https://${BUCKET}.s3.amazonaws.com/build-stats/${SHA}/${GITHUB_RUN_ID}/cargo-timing.html
-          aws s3 cp --only-show-errors ./target/cargo-timings/cargo-timing.html "s3://${BUCKET}/build-stats/${SHA}/${GITHUB_RUN_ID}/"
-          echo "report-url=${REPORT_URL}" >> $GITHUB_OUTPUT
-
-      - name: Publish build stats report
-        uses: actions/github-script@v6
-        env:
-          REPORT_URL: ${{ steps.upload-stats.outputs.report-url }}
-          SHA: ${{ github.event.pull_request.head.sha || github.sha }}
+        uses: actions/upload-artifact@v3
         with:
-          script: |
-            const { REPORT_URL, SHA } = process.env
-
-            await github.rest.repos.createCommitStatus({
-              owner: context.repo.owner,
-              repo: context.repo.repo,
-              sha: `${SHA}`,
-              state: 'success',
-              target_url: `${REPORT_URL}`,
-              context: `Build stats (release)`,
-            })
+          name: neon-${{ runner.os }}-release-build-stats
+          path: ./target/cargo-timings/

.github/workflows/release.yml

@@ -1,34 +0,0 @@
-name: Create Release Branch
-
-on:
-  schedule:
-    - cron: '0 10 * * 2'
-
-jobs:
-  create_release_branch:
-    runs-on: [ubuntu-latest]
-
-    steps:
-    - name: Check out code
-      uses: actions/checkout@v3
-      with:
-        ref: main
-
-    - name: Get current date
-      id: date
-      run: echo "date=$(date +'%Y-%m-%d')" >> $GITHUB_OUTPUT
-
-    - name: Create release branch
-      run: git checkout -b releases/${{ steps.date.outputs.date }}
-
-    - name: Push new branch
-      run: git push origin releases/${{ steps.date.outputs.date }}
-
-    - name: Create pull request into release
-      uses: thomaseizinger/create-pull-request@e3972219c86a56550fb70708d96800d8e24ba862 # 1.3.0
-      with:
-        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        head: releases/${{ steps.date.outputs.date }}
-        base: release
-        title: Release ${{ steps.date.outputs.date }}
-        team_reviewers: release

Cargo.toml

@@ -7,7 +7,6 @@ members = [
     "safekeeper",
     "storage_broker",
     "workspace_hack",
-    "trace",
     "libs/*",
 ]
 
@@ -32,15 +31,12 @@ bstr = "1.0"
 byteorder = "1.4"
 bytes = "1.0"
 chrono = { version = "0.4", default-features = false, features = ["clock"] }
-clap = { version = "4.0", features = ["derive"] }
+clap = "4.0"
 close_fds = "0.3.2"
 comfy-table = "6.1"
 const_format = "0.2"
 crc32c = "0.6"
 crossbeam-utils = "0.8.5"
-either = "1.8"
-enum-map = "2.4.2"
-enumset = "1.0.12"
 fail = "0.5.0"
 fs2 = "0.4.3"
 futures = "0.3"
@@ -48,7 +44,6 @@ futures-core = "0.3"
 futures-util = "0.3"
 git-version = "0.3"
 hashbrown = "0.13"
-hashlink = "0.8.1"
 hex = "0.4"
 hex-literal = "0.3"
 hmac = "0.12.1"
@@ -64,12 +59,8 @@ md5 = "0.7.0"
 memoffset = "0.8"
 nix = "0.26"
 notify = "5.0.0"
-num_cpus = "1.15"
 num-traits = "0.2.15"
 once_cell = "1.13"
-opentelemetry = "0.18.0"
-opentelemetry-otlp = { version = "0.11.0", default_features=false, features = ["http-proto", "trace", "http", "reqwest-client"] }
-opentelemetry-semantic-conventions = "0.10.0"
 parking_lot = "0.12"
 pin-project-lite = "0.2"
 prometheus = {version = "0.13", default_features=false, features = ["process"]} # removes protobuf dependency
@@ -77,8 +68,6 @@ prost = "0.11"
 rand = "0.8"
 regex = "1.4"
 reqwest = { version = "0.11", default-features = false, features = ["rustls-tls"] }
-reqwest-tracing = { version = "0.4.0", features = ["opentelemetry_0_18"] }
-reqwest-middleware = "0.2.0"
 routerify = "3"
 rpds = "0.12.0"
 rustls = "0.20"
@@ -95,7 +84,6 @@ socket2 = "0.4.4"
 strum = "0.24"
 strum_macros = "0.24"
 svg_fmt = "0.4.1"
-sync_wrapper = "0.1.2"
 tar = "0.4"
 thiserror = "1.0"
 tls-listener = { version = "0.6", features = ["rustls", "hyper-h1"] }
@@ -108,7 +96,6 @@ toml = "0.5"
 toml_edit = { version = "0.17", features = ["easy"] }
 tonic = {version = "0.8", features = ["tls", "tls-roots"]}
 tracing = "0.1"
-tracing-opentelemetry = "0.18.0"
 tracing-subscriber = { version = "0.3", features = ["env-filter"] }
 url = "2.2"
 uuid = { version = "1.2", features = ["v4", "serde"] }
@@ -127,14 +114,10 @@ postgres-types = { git = "https://github.com/neondatabase/rust-postgres.git", re
 tokio-postgres = { git = "https://github.com/neondatabase/rust-postgres.git", rev="43e6db254a97fdecbce33d8bc0890accfd74495e" }
 tokio-tar = { git = "https://github.com/neondatabase/tokio-tar.git", rev="404df61437de0feef49ba2ccdbdd94eb8ad6e142" }
 
-## Other git libraries
-heapless = { default-features=false, features=[], git = "https://github.com/japaric/heapless.git", rev = "644653bf3b831c6bb4963be2de24804acf5e5001" } # upstream release pending
-
 ## Local libraries
 consumption_metrics = { version = "0.1", path = "./libs/consumption_metrics/" }
 metrics = { version = "0.1", path = "./libs/metrics/" }
 pageserver_api = { version = "0.1", path = "./libs/pageserver_api/" }
-postgres_backend = { version = "0.1", path = "./libs/postgres_backend/" }
 postgres_connection = { version = "0.1", path = "./libs/postgres_connection/" }
 postgres_ffi = { version = "0.1", path = "./libs/postgres_ffi/" }
 pq_proto = { version = "0.1", path = "./libs/pq_proto/" }
@@ -142,7 +125,6 @@ remote_storage = { version = "0.1", path = "./libs/remote_storage/" }
 safekeeper_api = { version = "0.1", path = "./libs/safekeeper_api" }
 storage_broker = { version = "0.1", path = "./storage_broker/" } # Note: main broker code is inside the binary crate, so linking with the library shouldn't be heavy.
 tenant_size_model = { version = "0.1", path = "./libs/tenant_size_model/" }
-tracing-utils = { version = "0.1", path = "./libs/tracing-utils/" }
 utils = { version = "0.1", path = "./libs/utils/" }
 
 ## Common library dependency
@@ -152,7 +134,7 @@ workspace_hack = { version = "0.1", path = "./workspace_hack/" }
 criterion = "0.4"
 rcgen = "0.10"
 rstest = "0.16"
-tempfile = "3.4"
+tempfile = "3.2"
 tonic-build = "0.8"
 
 # This is only needed for proxy's tests.

Dockerfile

@@ -39,7 +39,7 @@ ARG CACHEPOT_BUCKET=neon-github-dev
 
 COPY --from=pg-build /home/nonroot/pg_install/v14/include/postgresql/server pg_install/v14/include/postgresql/server
 COPY --from=pg-build /home/nonroot/pg_install/v15/include/postgresql/server pg_install/v15/include/postgresql/server
-COPY --chown=nonroot . .
+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.

Dockerfile.compute-node

@@ -1,493 +0,0 @@
-ARG PG_VERSION
-ARG REPOSITORY=369495373322.dkr.ecr.eu-central-1.amazonaws.com
-ARG IMAGE=rust
-ARG TAG=pinned
-
-#########################################################################################
-#
-# Layer "build-deps"
-#
-#########################################################################################
-FROM debian:bullseye-slim AS build-deps
-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 libssl-dev \
-    libicu-dev libxslt1-dev
-
-#########################################################################################
-#
-# Layer "pg-build"
-# Build Postgres from the neon postgres repository.
-#
-#########################################################################################
-FROM build-deps AS pg-build
-ARG PG_VERSION
-COPY vendor/postgres-${PG_VERSION} postgres
-RUN cd postgres && \
-    ./configure CFLAGS='-O2 -g3' --enable-debug --with-openssl --with-uuid=ossp --with-icu \
-    --with-libxml --with-libxslt && \
-    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 && \
-    # Enable some of contrib extensions
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/autoinc.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/bloom.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/earthdistance.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/insert_username.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/intagg.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/moddatetime.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/pgrowlocks.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/pgstattuple.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/refint.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/xml2.control
-
-#########################################################################################
-#
-# Layer "postgis-build"
-# Build PostGIS from the upstream PostGIS mirror.
-#
-#########################################################################################
-FROM build-deps AS postgis-build
-COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
-RUN apt update && \
-    apt install -y cmake gdal-bin libboost-dev libboost-thread-dev libboost-filesystem-dev \
-    libboost-system-dev libboost-iostreams-dev libboost-program-options-dev libboost-timer-dev \
-    libcgal-dev libgdal-dev libgmp-dev libmpfr-dev libopenscenegraph-dev libprotobuf-c-dev \
-    protobuf-c-compiler xsltproc
-
-# SFCGAL > 1.3 requires CGAL > 5.2, Bullseye's libcgal-dev is 5.2
-RUN wget https://gitlab.com/Oslandia/SFCGAL/-/archive/v1.3.10/SFCGAL-v1.3.10.tar.gz -O SFCGAL.tar.gz && \
-    mkdir sfcgal-src && cd sfcgal-src && tar xvzf ../SFCGAL.tar.gz --strip-components=1 -C . && \
-    cmake . && make -j $(getconf _NPROCESSORS_ONLN) && \
-    DESTDIR=/sfcgal make install -j $(getconf _NPROCESSORS_ONLN) && \
-    make clean && cp -R /sfcgal/* /
-
-ENV PATH "/usr/local/pgsql/bin:$PATH"
-
-RUN wget https://download.osgeo.org/postgis/source/postgis-3.3.2.tar.gz -O postgis.tar.gz && \
-    mkdir postgis-src && cd postgis-src && tar xvzf ../postgis.tar.gz --strip-components=1 -C . && \
-    ./autogen.sh && \
-    ./configure --with-sfcgal=/usr/local/bin/sfcgal-config && \
-    make -j $(getconf _NPROCESSORS_ONLN) install && \
-    cd extensions/postgis && \
-    make clean && \
-    make -j $(getconf _NPROCESSORS_ONLN) install && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/postgis.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/postgis_raster.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/postgis_sfcgal.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/postgis_tiger_geocoder.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/postgis_topology.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/address_standardizer.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/address_standardizer_data_us.control
-
-RUN wget https://github.com/pgRouting/pgrouting/archive/v3.4.2.tar.gz -O pgrouting.tar.gz && \
-    mkdir pgrouting-src && cd pgrouting-src && tar xvzf ../pgrouting.tar.gz --strip-components=1 -C . && \
-    mkdir build && \
-    cd build && \
-    cmake .. && \
-    make -j $(getconf _NPROCESSORS_ONLN) && \
-    make -j $(getconf _NPROCESSORS_ONLN) install && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/pgrouting.control
-
-#########################################################################################
-#
-# Layer "plv8-build"
-# Build plv8
-#
-#########################################################################################
-FROM build-deps AS plv8-build
-COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
-RUN apt update && \
-    apt install -y ninja-build python3-dev libncurses5 binutils clang
-
-RUN wget https://github.com/plv8/plv8/archive/refs/tags/v3.1.5.tar.gz -O plv8.tar.gz && \
-    mkdir plv8-src && cd plv8-src && tar xvzf ../plv8.tar.gz --strip-components=1 -C . && \
-    export PATH="/usr/local/pgsql/bin:$PATH" && \
-    make DOCKER=1 -j $(getconf _NPROCESSORS_ONLN) install && \
-    rm -rf /plv8-* && \
-    find /usr/local/pgsql/ -name "plv8-*.so" | xargs strip && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/plv8.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/plcoffee.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/plls.control
-
-#########################################################################################
-#
-# Layer "h3-pg-build"
-# Build h3_pg
-#
-#########################################################################################
-FROM build-deps AS h3-pg-build
-COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
-
-# packaged cmake is too old
-RUN wget https://github.com/Kitware/CMake/releases/download/v3.24.2/cmake-3.24.2-linux-x86_64.sh \
-      -q -O /tmp/cmake-install.sh \
-      && chmod u+x /tmp/cmake-install.sh \
-      && /tmp/cmake-install.sh --skip-license --prefix=/usr/local/ \
-      && rm /tmp/cmake-install.sh
-
-RUN wget https://github.com/uber/h3/archive/refs/tags/v4.1.0.tar.gz -O h3.tar.gz && \
-    mkdir h3-src && cd h3-src && tar xvzf ../h3.tar.gz --strip-components=1 -C . && \
-    mkdir build && cd build && \
-    cmake .. -DCMAKE_BUILD_TYPE=Release && \
-    make -j $(getconf _NPROCESSORS_ONLN) && \
-    DESTDIR=/h3 make install && \
-    cp -R /h3/usr / && \
-    rm -rf build
-
-RUN wget https://github.com/zachasme/h3-pg/archive/refs/tags/v4.1.2.tar.gz -O h3-pg.tar.gz && \
-    mkdir h3-pg-src && cd h3-pg-src && tar xvzf ../h3-pg.tar.gz --strip-components=1 -C . && \
-    export PATH="/usr/local/pgsql/bin:$PATH" && \
-    make -j $(getconf _NPROCESSORS_ONLN) && \
-    make -j $(getconf _NPROCESSORS_ONLN) install && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/h3.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/h3_postgis.control
-
-#########################################################################################
-#
-# Layer "unit-pg-build"
-# compile unit extension
-#
-#########################################################################################
-FROM build-deps AS unit-pg-build
-COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
-
-RUN wget https://github.com/df7cb/postgresql-unit/archive/refs/tags/7.7.tar.gz -O postgresql-unit.tar.gz && \
-    mkdir postgresql-unit-src && cd postgresql-unit-src && tar xvzf ../postgresql-unit.tar.gz --strip-components=1 -C . && \
-    make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
-    make -j $(getconf _NPROCESSORS_ONLN) install PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
-    # unit extension's "create extension" script relies on absolute install path to fill some reference tables.
-    # We move the extension from '/usr/local/pgsql/' to '/usr/local/'  after it is build. So we need to adjust the path.
-    # This one-liner removes pgsql/ part of the path.
-    # NOTE: Other extensions that rely on MODULEDIR variable after building phase will need the same fix.
-    find /usr/local/pgsql/share/extension/ -name "unit*.sql" -print0 | xargs -0 sed -i "s|pgsql/||g" && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/unit.control
-
-#########################################################################################
-#
-# Layer "vector-pg-build"
-# compile pgvector extension
-#
-#########################################################################################
-FROM build-deps AS vector-pg-build
-COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
-
-RUN wget https://github.com/pgvector/pgvector/archive/refs/tags/v0.4.0.tar.gz -O pgvector.tar.gz && \
-    mkdir pgvector-src && cd pgvector-src && tar xvzf ../pgvector.tar.gz --strip-components=1 -C . && \
-    make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
-    make -j $(getconf _NPROCESSORS_ONLN) install PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/vector.control
-
-#########################################################################################
-#
-# Layer "pgjwt-pg-build"
-# compile pgjwt extension
-#
-#########################################################################################
-FROM build-deps AS pgjwt-pg-build
-COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
-
-# 9742dab1b2f297ad3811120db7b21451bca2d3c9 made on 13/11/2021
-RUN wget https://github.com/michelp/pgjwt/archive/9742dab1b2f297ad3811120db7b21451bca2d3c9.tar.gz -O pgjwt.tar.gz && \
-    mkdir pgjwt-src && cd pgjwt-src && tar xvzf ../pgjwt.tar.gz --strip-components=1 -C . && \
-    make -j $(getconf _NPROCESSORS_ONLN) install PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/pgjwt.control
-
-#########################################################################################
-#
-# Layer "hypopg-pg-build"
-# compile hypopg extension
-#
-#########################################################################################
-FROM build-deps AS hypopg-pg-build
-COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
-
-RUN wget https://github.com/HypoPG/hypopg/archive/refs/tags/1.3.1.tar.gz -O hypopg.tar.gz && \
-    mkdir hypopg-src && cd hypopg-src && tar xvzf ../hypopg.tar.gz --strip-components=1 -C . && \
-    make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
-    make -j $(getconf _NPROCESSORS_ONLN) install PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/hypopg.control
-
-#########################################################################################
-#
-# Layer "pg-hashids-pg-build"
-# compile pg_hashids extension
-#
-#########################################################################################
-FROM build-deps AS pg-hashids-pg-build
-COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
-
-RUN wget https://github.com/iCyberon/pg_hashids/archive/refs/tags/v1.2.1.tar.gz -O pg_hashids.tar.gz && \
-    mkdir pg_hashids-src && cd pg_hashids-src && tar xvzf ../pg_hashids.tar.gz --strip-components=1 -C . && \
-    make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config USE_PGXS=1 && \
-    make -j $(getconf _NPROCESSORS_ONLN) install PG_CONFIG=/usr/local/pgsql/bin/pg_config USE_PGXS=1 && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/pg_hashids.control
-
-#########################################################################################
-#
-# Layer "rum-pg-build"
-# compile rum extension
-#
-#########################################################################################
-FROM build-deps AS rum-pg-build
-COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
-
-RUN wget https://github.com/postgrespro/rum/archive/refs/tags/1.3.13.tar.gz -O rum.tar.gz && \
-    mkdir rum-src && cd rum-src && tar xvzf ../rum.tar.gz --strip-components=1 -C . && \
-    make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config USE_PGXS=1 && \
-    make -j $(getconf _NPROCESSORS_ONLN) install PG_CONFIG=/usr/local/pgsql/bin/pg_config USE_PGXS=1 && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/rum.control
-
-#########################################################################################
-#
-# Layer "pgtap-pg-build"
-# compile pgTAP extension
-#
-#########################################################################################
-FROM build-deps AS pgtap-pg-build
-COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
-
-RUN wget https://github.com/theory/pgtap/archive/refs/tags/v1.2.0.tar.gz -O pgtap.tar.gz && \
-    mkdir pgtap-src && cd pgtap-src && tar xvzf ../pgtap.tar.gz --strip-components=1 -C . && \
-    make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
-    make -j $(getconf _NPROCESSORS_ONLN) install PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/pgtap.control
-
-#########################################################################################
-#
-# Layer "prefix-pg-build"
-# compile Prefix extension
-#
-#########################################################################################
-FROM build-deps AS prefix-pg-build
-COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
-
-RUN wget https://github.com/dimitri/prefix/archive/refs/tags/v1.2.9.tar.gz -O prefix.tar.gz && \
-    mkdir prefix-src && cd prefix-src && tar xvzf ../prefix.tar.gz --strip-components=1 -C . && \
-    make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
-    make -j $(getconf _NPROCESSORS_ONLN) install PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/prefix.control
-
-#########################################################################################
-#
-# Layer "hll-pg-build"
-# compile hll extension
-#
-#########################################################################################
-FROM build-deps AS hll-pg-build
-COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
-
-RUN wget https://github.com/citusdata/postgresql-hll/archive/refs/tags/v2.17.tar.gz -O hll.tar.gz && \
-    mkdir hll-src && cd hll-src && tar xvzf ../hll.tar.gz --strip-components=1 -C . && \
-    make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
-    make -j $(getconf _NPROCESSORS_ONLN) install PG_CONFIG=/usr/local/pgsql/bin/pg_config && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/hll.control
-
-#########################################################################################
-#
-# Layer "plpgsql-check-pg-build"
-# compile plpgsql_check extension
-#
-#########################################################################################
-FROM build-deps AS plpgsql-check-pg-build
-COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
-
-RUN wget https://github.com/okbob/plpgsql_check/archive/refs/tags/v2.3.2.tar.gz -O plpgsql_check.tar.gz && \
-    mkdir plpgsql_check-src && cd plpgsql_check-src && tar xvzf ../plpgsql_check.tar.gz --strip-components=1 -C . && \
-    make -j $(getconf _NPROCESSORS_ONLN) PG_CONFIG=/usr/local/pgsql/bin/pg_config USE_PGXS=1 && \
-    make -j $(getconf _NPROCESSORS_ONLN) install PG_CONFIG=/usr/local/pgsql/bin/pg_config USE_PGXS=1 && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/plpgsql_check.control
-
-#########################################################################################
-# 
-# Layer "rust extensions"
-# This layer is used to build `pgx` deps
-#
-#########################################################################################
-FROM build-deps AS rust-extensions-build
-COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
-
-RUN apt-get update && \
-    apt-get install -y curl libclang-dev cmake && \
-    useradd -ms /bin/bash nonroot -b /home
-
-ENV HOME=/home/nonroot
-ENV PATH="/home/nonroot/.cargo/bin:/usr/local/pgsql/bin/:$PATH"
-USER nonroot
-WORKDIR /home/nonroot
-ARG PG_VERSION
-
-RUN curl -sSO https://static.rust-lang.org/rustup/dist/$(uname -m)-unknown-linux-gnu/rustup-init && \
-    chmod +x rustup-init && \
-    ./rustup-init -y --no-modify-path --profile minimal --default-toolchain stable && \
-    rm rustup-init && \
-    cargo install --locked --version 0.7.3 cargo-pgx && \
-    /bin/bash -c 'cargo pgx init --pg${PG_VERSION:1}=/usr/local/pgsql/bin/pg_config'
-
-USER root
-
-#########################################################################################
-# 
-# Layer "pg-jsonschema-pg-build"
-# Compile "pg_jsonschema" extension
-#
-#########################################################################################
-
-FROM rust-extensions-build AS pg-jsonschema-pg-build
-
-# there is no release tag yet, but we need it due to the superuser fix in the control file
-RUN wget https://github.com/supabase/pg_jsonschema/archive/caeab60d70b2fd3ae421ec66466a3abbb37b7ee6.tar.gz -O pg_jsonschema.tar.gz && \
-    mkdir pg_jsonschema-src && cd pg_jsonschema-src && tar xvzf ../pg_jsonschema.tar.gz --strip-components=1 -C . && \
-    sed -i 's/pgx = "0.7.1"/pgx = { version = "0.7.3", features = [ "unsafe-postgres" ] }/g' Cargo.toml && \
-    cargo pgx install --release && \
-    echo "trusted = true" >> /usr/local/pgsql/share/extension/pg_jsonschema.control
-
-#########################################################################################
-# 
-# Layer "pg-graphql-pg-build"
-# Compile "pg_graphql" extension
-#
-#########################################################################################
-
-FROM rust-extensions-build AS pg-graphql-pg-build
-
-# Currently pgx version bump to >= 0.7.2  causes "call to unsafe function" compliation errors in
-# pgx-contrib-spiext. There is a branch that removes that dependency, so use it. It is on the
-# same 1.1 version we've used before.
-RUN git clone -b remove-pgx-contrib-spiext --single-branch https://github.com/yrashk/pg_graphql && \
-    cd pg_graphql && \
-    sed -i 's/pgx = "~0.7.1"/pgx = { version = "0.7.3", features = [ "unsafe-postgres" ] }/g' Cargo.toml && \
-    sed -i 's/pgx-tests = "~0.7.1"/pgx-tests = "0.7.3"/g' Cargo.toml && \
-    cargo pgx install --release && \
-    # it's needed to enable extension because it uses untrusted C language
-    sed -i 's/superuser = false/superuser = true/g' /usr/local/pgsql/share/extension/pg_graphql.control && \
-    echo "trusted = true" >> /usr/local/pgsql/share/extension/pg_graphql.control
-
-#########################################################################################
-#
-# Layer "pg-tiktoken-build"
-# Compile "pg_tiktoken" extension
-#
-#########################################################################################
-
-FROM rust-extensions-build AS pg-tiktoken-pg-build
-
-RUN git clone --depth=1 --single-branch https://github.com/kelvich/pg_tiktoken && \
-    cd pg_tiktoken && \
-    cargo pgx install --release && \
-    echo "trusted = true" >> /usr/local/pgsql/share/extension/pg_tiktoken.control
-
-#########################################################################################
-#
-# Layer "neon-pg-ext-build"
-# compile neon extensions
-#
-#########################################################################################
-FROM build-deps AS neon-pg-ext-build
-COPY --from=postgis-build /usr/local/pgsql/ /usr/local/pgsql/
-COPY --from=postgis-build /sfcgal/* /
-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 --from=unit-pg-build /usr/local/pgsql/ /usr/local/pgsql/
-COPY --from=vector-pg-build /usr/local/pgsql/ /usr/local/pgsql/
-COPY --from=pgjwt-pg-build /usr/local/pgsql/ /usr/local/pgsql/
-COPY --from=pg-jsonschema-pg-build /usr/local/pgsql/ /usr/local/pgsql/
-COPY --from=pg-graphql-pg-build /usr/local/pgsql/ /usr/local/pgsql/
-COPY --from=pg-tiktoken-pg-build /usr/local/pgsql/ /usr/local/pgsql/
-COPY --from=hypopg-pg-build /usr/local/pgsql/ /usr/local/pgsql/
-COPY --from=pg-hashids-pg-build /usr/local/pgsql/ /usr/local/pgsql/
-COPY --from=rum-pg-build /usr/local/pgsql/ /usr/local/pgsql/
-COPY --from=pgtap-pg-build /usr/local/pgsql/ /usr/local/pgsql/
-COPY --from=prefix-pg-build /usr/local/pgsql/ /usr/local/pgsql/
-COPY --from=hll-pg-build /usr/local/pgsql/ /usr/local/pgsql/
-COPY --from=plpgsql-check-pg-build /usr/local/pgsql/ /usr/local/pgsql/
-COPY pgxn/ pgxn/
-
-RUN make -j $(getconf _NPROCESSORS_ONLN) \
-        PG_CONFIG=/usr/local/pgsql/bin/pg_config \
-        -C pgxn/neon \
-        -s install && \
-    make -j $(getconf _NPROCESSORS_ONLN) \
-        PG_CONFIG=/usr/local/pgsql/bin/pg_config \
-        -C pgxn/neon_utils \
-        -s install
-
-#########################################################################################
-#
-# Compile and run the Neon-specific `compute_ctl` binary
-#
-#########################################################################################
-FROM $REPOSITORY/$IMAGE:$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
-
-# Remove binaries from /bin/ that we won't use (or would manually copy & install otherwise)
-RUN cd /usr/local/pgsql/bin && rm ecpg raster2pgsql shp2pgsql pgtopo_export pgtopo_import pgsql2shp
-
-# Remove headers that we won't need anymore - we've completed installation of all extensions
-RUN rm -r /usr/local/pgsql/include
-
-# Remove static postgresql libraries - all compilation is finished, so we
-# can now remove these files - they must be included in other binaries by now
-# 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 && \
-    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 && \
-    echo '/usr/local/lib' >> /etc/ld.so.conf && /sbin/ldconfig && \
-    # create folder for file cache
-    mkdir -p -m 777 /neon/cache
-
-COPY --from=postgres-cleanup-layer --chown=postgres /usr/local/pgsql /usr/local
-COPY --from=compute-tools --chown=postgres /home/nonroot/target/release-line-debug-size-lto/compute_ctl /usr/local/bin/compute_ctl
-
-# Install:
-# libreadline8 for psql
-# libicu67, locales for collations (including ICU and plpgsql_check)
-# libossp-uuid16 for extension ossp-uuid
-# libgeos, libgdal, libsfcgal1, libproj and libprotobuf-c1 for PostGIS
-# libxml2, libxslt1.1 for xml2
-RUN apt update &&  \
-    apt install --no-install-recommends -y \
-        locales \
-        libicu67 \
-        libreadline8 \
-        libossp-uuid16 \
-        libgeos-c1v5 \
-        libgdal28 \
-        libproj19 \
-        libprotobuf-c1 \
-        libsfcgal1 \
-        libxml2 \
-        libxslt1.1 \
-        gdb && \
-    rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/* && \
-    localedef -i en_US -c -f UTF-8 -A /usr/share/locale/locale.alias en_US.UTF-8
-
-ENV LANG en_US.utf8
-USER postgres
-ENTRYPOINT ["/usr/local/bin/compute_ctl"]

Dockerfile.compute-node-v14

@@ -0,0 +1,220 @@
+#
+# 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 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 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-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 $(getconf _NPROCESSORS_ONLN) -s -C src/interfaces/libpq install && \
+    # Enable some of contrib extensions
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/bloom.control && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/pgrowlocks.control && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/intagg.control && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/pgstattuple.control && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/earthdistance.control
+
+#########################################################################################
+#
+# Layer "postgis-build"
+# Build PostGIS from the upstream PostGIS mirror.
+#
+#########################################################################################
+FROM build-deps AS postgis-build
+COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
+RUN apt update && \
+    apt install -y gdal-bin libgdal-dev libprotobuf-c-dev protobuf-c-compiler xsltproc
+
+RUN wget https://download.osgeo.org/postgis/source/postgis-3.3.1.tar.gz && \
+    tar xvzf postgis-3.3.1.tar.gz && \
+    cd postgis-3.3.1 && \
+    ./autogen.sh && \
+    export PATH="/usr/local/pgsql/bin:$PATH" && \
+    ./configure && \
+    make -j $(getconf _NPROCESSORS_ONLN) install && \
+    cd extensions/postgis && \
+    make clean && \
+    make -j $(getconf _NPROCESSORS_ONLN) install && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/postgis.control && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/postgis_raster.control && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/postgis_tiger_geocoder.control && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/postgis_topology.control && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/address_standardizer.control && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/address_standardizer_data_us.control
+
+#########################################################################################
+#
+# Layer "plv8-build"
+# Build plv8
+#
+#########################################################################################
+FROM build-deps AS plv8-build
+COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
+RUN apt update && \
+    apt install -y ninja-build python3-dev libc++-dev libc++abi-dev libncurses5 binutils
+
+# 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" && \
+    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 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  && \
+    cd h3-4.0.1 && \
+    mkdir build && \
+    cd build && \
+    cmake .. -DCMAKE_BUILD_TYPE=Release && \
+    make -j $(getconf _NPROCESSORS_ONLN) && \
+    DESTDIR=/h3 make install && \
+    cp -R /h3/usr / && \
+    rm -rf build
+
+RUN wget https://github.com/zachasme/h3-pg/archive/refs/tags/v4.0.1.tar.gz -O h3-pg.tgz && \
+    tar xvzf h3-pg.tgz && \
+    cd h3-pg-4.0.1 && \
+    export PATH="/usr/local/pgsql/bin:$PATH" && \
+    make -j $(getconf _NPROCESSORS_ONLN) && \
+    make -j $(getconf _NPROCESSORS_ONLN) install && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/h3.control && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/h3_postgis.control
+
+#########################################################################################
+#
+# Layer "neon-pg-ext-build"
+# compile neon extensions
+#
+#########################################################################################
+FROM build-deps AS neon-pg-ext-build
+COPY --from=postgis-build /usr/local/pgsql/ /usr/local/pgsql/
+COPY --from=plv8-build /usr/local/pgsql/ /usr/local/pgsql/
+COPY --from=h3-pg-build /usr/local/pgsql/ /usr/local/pgsql/
+COPY --from=h3-pg-build /h3/usr /
+COPY pgxn/ pgxn/
+
+RUN make -j $(getconf _NPROCESSORS_ONLN) \
+        PG_CONFIG=/usr/local/pgsql/bin/pg_config \
+        -C pgxn/neon \
+        -s install
+
+#########################################################################################
+#
+# Compile and run the Neon-specific `compute_ctl` binary
+#
+#########################################################################################
+FROM 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:$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
+
+# Remove binaries from /bin/ that we won't use (or would manually copy & install otherwise)
+RUN cd /usr/local/pgsql/bin && rm ecpg raster2pgsql shp2pgsql pgtopo_export pgtopo_import pgsql2shp
+
+# Remove headers that we won't need anymore - we've completed installation of all extensions
+RUN rm -r /usr/local/pgsql/include
+
+# Remove static postgresql libraries - all compilation is finished, so we
+# can now remove these files - they must be included in other binaries by now
+# 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 && \
+    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 && \
+    echo '/usr/local/lib' >> /etc/ld.so.conf && /sbin/ldconfig
+
+COPY --from=postgres-cleanup-layer --chown=postgres /usr/local/pgsql /usr/local
+COPY --from=compute-tools --chown=postgres /home/nonroot/target/release-line-debug-size-lto/compute_ctl /usr/local/bin/compute_ctl
+
+# Install:
+# libreadline8 for psql
+# libossp-uuid16 for extension ossp-uuid
+# libgeos, libgdal, libproj and libprotobuf-c1 for PostGIS
+RUN apt update &&  \
+    apt install --no-install-recommends -y \
+        libreadline8 \
+        libossp-uuid16 \
+        libgeos-c1v5 \
+        libgdal28 \
+        libproj19 \
+        libprotobuf-c1 \
+        gdb && \
+    rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/*
+
+USER postgres
+ENTRYPOINT ["/usr/local/bin/compute_ctl"]

Dockerfile.compute-node-v15

@@ -0,0 +1,220 @@
+#
+# This file is identical to the Dockerfile.compute-node-v14 file
+# except for the version of Postgres that is built.
+#
+
+ARG TAG=pinned
+
+#########################################################################################
+#
+# Layer "build-deps"
+#
+#########################################################################################
+FROM debian:bullseye-slim AS build-deps
+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 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-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 $(getconf _NPROCESSORS_ONLN) -s -C src/interfaces/libpq install && \
+    # Enable some of contrib extensions
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/bloom.control && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/pgrowlocks.control && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/intagg.control && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/pgstattuple.control && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/earthdistance.control
+
+#########################################################################################
+#
+# Layer "postgis-build"
+# Build PostGIS from the upstream PostGIS mirror.
+#
+#########################################################################################
+FROM build-deps AS postgis-build
+COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
+RUN apt update && \
+    apt install -y gdal-bin libgdal-dev libprotobuf-c-dev protobuf-c-compiler xsltproc
+
+RUN wget https://download.osgeo.org/postgis/source/postgis-3.3.1.tar.gz && \
+    tar xvzf postgis-3.3.1.tar.gz && \
+    cd postgis-3.3.1 && \
+    ./autogen.sh && \
+    export PATH="/usr/local/pgsql/bin:$PATH" && \
+    ./configure && \
+    make -j $(getconf _NPROCESSORS_ONLN) install && \
+    cd extensions/postgis && \
+    make clean && \
+    make -j $(getconf _NPROCESSORS_ONLN) install && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/postgis.control && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/postgis_raster.control && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/postgis_tiger_geocoder.control && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/postgis_topology.control && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/address_standardizer.control && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/address_standardizer_data_us.control
+
+#########################################################################################
+#
+# Layer "plv8-build"
+# Build plv8
+#
+#########################################################################################
+FROM build-deps AS plv8-build
+COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
+RUN apt update && \
+    apt install -y ninja-build python3-dev libc++-dev libc++abi-dev libncurses5 binutils
+
+# 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" && \
+    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 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  && \
+    cd h3-4.0.1 && \
+    mkdir build && \
+    cd build && \
+    cmake .. -DCMAKE_BUILD_TYPE=Release && \
+    make -j $(getconf _NPROCESSORS_ONLN) && \
+    DESTDIR=/h3 make install && \
+    cp -R /h3/usr / && \
+    rm -rf build
+
+RUN wget https://github.com/zachasme/h3-pg/archive/refs/tags/v4.0.1.tar.gz -O h3-pg.tgz && \
+    tar xvzf h3-pg.tgz && \
+    cd h3-pg-4.0.1 && \
+    export PATH="/usr/local/pgsql/bin:$PATH" && \
+    make -j $(getconf _NPROCESSORS_ONLN) && \
+    make -j $(getconf _NPROCESSORS_ONLN) install && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/h3.control && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/h3_postgis.control
+
+#########################################################################################
+#
+# Layer "neon-pg-ext-build"
+# compile neon extensions
+#
+#########################################################################################
+FROM build-deps AS neon-pg-ext-build
+COPY --from=postgis-build /usr/local/pgsql/ /usr/local/pgsql/
+COPY --from=plv8-build /usr/local/pgsql/ /usr/local/pgsql/
+COPY --from=h3-pg-build /usr/local/pgsql/ /usr/local/pgsql/
+COPY --from=h3-pg-build /h3/usr /
+COPY pgxn/ pgxn/
+
+RUN make -j $(getconf _NPROCESSORS_ONLN) \
+        PG_CONFIG=/usr/local/pgsql/bin/pg_config \
+        -C pgxn/neon \
+        -s install
+
+#########################################################################################
+#
+# Compile and run the Neon-specific `compute_ctl` binary
+#
+#########################################################################################
+FROM 369495373322.dkr.ecr.eu-central-1.amazonaws.com/rust:$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
+
+# Remove binaries from /bin/ that we won't use (or would manually copy & install otherwise)
+RUN cd /usr/local/pgsql/bin && rm ecpg raster2pgsql shp2pgsql pgtopo_export pgtopo_import pgsql2shp
+
+# Remove headers that we won't need anymore - we've completed installation of all extensions
+RUN rm -r /usr/local/pgsql/include
+
+# Remove static postgresql libraries - all compilation is finished, so we
+# can now remove these files - they must be included in other binaries by now
+# 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 && \
+    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 && \
+    echo '/usr/local/lib' >> /etc/ld.so.conf && /sbin/ldconfig
+
+COPY --from=postgres-cleanup-layer --chown=postgres /usr/local/pgsql /usr/local
+COPY --from=compute-tools --chown=postgres /home/nonroot/target/release-line-debug-size-lto/compute_ctl /usr/local/bin/compute_ctl
+
+# Install:
+# libreadline8 for psql
+# libossp-uuid16 for extension ossp-uuid
+# libgeos, libgdal, libproj and libprotobuf-c1 for PostGIS
+RUN apt update &&  \
+    apt install --no-install-recommends -y \
+        libreadline8 \
+        libossp-uuid16 \
+        libgeos-c1v5 \
+        libgdal28 \
+        libproj19 \
+        libprotobuf-c1 \
+        gdb && \
+    rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/*
+
+USER postgres
+ENTRYPOINT ["/usr/local/bin/compute_ctl"]

Dockerfile.vm-compute-node

@@ -1,70 +0,0 @@
-# Note: this file *mostly* just builds on Dockerfile.compute-node
-
-ARG SRC_IMAGE
-ARG VM_INFORMANT_VERSION=v0.1.14
-# on libcgroup update, make sure to check bootstrap.sh for changes
-ARG LIBCGROUP_VERSION=v2.0.3
-
-# Pull VM informant, to copy from later
-FROM neondatabase/vm-informant:$VM_INFORMANT_VERSION as informant
-
-# Build cgroup-tools
-#
-# At time of writing (2023-03-14), debian bullseye has a version of cgroup-tools (technically
-# libcgroup) that doesn't support cgroup v2 (version 0.41-11). Unfortunately, the vm-informant
-# requires cgroup v2, so we'll build cgroup-tools ourselves.
-FROM debian:bullseye-slim as libcgroup-builder
-ARG LIBCGROUP_VERSION
-
-RUN set -exu \
-	&& apt update \
-	&& apt install --no-install-recommends -y \
-		git \
-		ca-certificates \
-		automake \
-		cmake \
-		make \
-		gcc \
-		byacc \
-		flex \
-		libtool \
-		libpam0g-dev \
-	&& git clone --depth 1 -b $LIBCGROUP_VERSION https://github.com/libcgroup/libcgroup \
-	&& INSTALL_DIR="/libcgroup-install" \
-	&& mkdir -p "$INSTALL_DIR/bin" "$INSTALL_DIR/include" \
-	&& cd libcgroup \
-	# extracted from bootstrap.sh, with modified flags:
-	&& (test -d m4 || mkdir m4) \
-	&& autoreconf -fi \
-	&& rm -rf autom4te.cache \
-	&& CFLAGS="-O3" ./configure --prefix="$INSTALL_DIR" --sysconfdir=/etc --localstatedir=/var --enable-opaque-hierarchy="name=systemd" \
-	# actually build the thing...
-	&& make install
-
-# Combine, starting from non-VM compute node image.
-FROM $SRC_IMAGE as base
-
-# Temporarily set user back to root so we can run adduser, set inittab
-USER root
-RUN adduser vm-informant --disabled-password --no-create-home
-
-RUN set -e \
-	&& rm -f /etc/inittab \
-	&& touch /etc/inittab
-
-RUN set -e \
-	&& echo "::sysinit:cgconfigparser -l /etc/cgconfig.conf -s 1664" >> /etc/inittab \
-	&& CONNSTR="dbname=neondb user=cloud_admin sslmode=disable" \
-	&& ARGS="--auto-restart --cgroup=neon-postgres --pgconnstr=\"$CONNSTR\"" \
-	&& echo "::respawn:su vm-informant -c '/usr/local/bin/vm-informant $ARGS'" >> /etc/inittab
-
-USER postgres
-
-ADD vm-cgconfig.conf /etc/cgconfig.conf
-COPY --from=informant /usr/bin/vm-informant /usr/local/bin/vm-informant
-
-COPY --from=libcgroup-builder /libcgroup-install/bin/* /usr/bin/
-COPY --from=libcgroup-builder /libcgroup-install/lib/* /usr/lib/
-COPY --from=libcgroup-builder /libcgroup-install/sbin/* /usr/sbin/
-
-ENTRYPOINT ["/usr/sbin/cgexec", "-g", "*:neon-postgres", "/usr/local/bin/compute_ctl"]

Makefile

@@ -133,26 +133,12 @@ neon-pg-ext-%: postgres-%
 	$(MAKE) PG_CONFIG=$(POSTGRES_INSTALL_DIR)/$*/bin/pg_config CFLAGS='$(PG_CFLAGS) $(COPT)' \
 		-C $(POSTGRES_INSTALL_DIR)/build/neon-test-utils-$* \
 		-f $(ROOT_PROJECT_DIR)/pgxn/neon_test_utils/Makefile install
-	+@echo "Compiling neon_utils $*"
-	mkdir -p $(POSTGRES_INSTALL_DIR)/build/neon-utils-$*
-	$(MAKE) PG_CONFIG=$(POSTGRES_INSTALL_DIR)/$*/bin/pg_config CFLAGS='$(PG_CFLAGS) $(COPT)' \
-		-C $(POSTGRES_INSTALL_DIR)/build/neon-utils-$* \
-		-f $(ROOT_PROJECT_DIR)/pgxn/neon_utils/Makefile install
 
 .PHONY: neon-pg-ext-clean-%
 neon-pg-ext-clean-%:
-	$(MAKE) PG_CONFIG=$(POSTGRES_INSTALL_DIR)/$*/bin/pg_config \
-	-C $(POSTGRES_INSTALL_DIR)/build/neon-$* \
-	-f $(ROOT_PROJECT_DIR)/pgxn/neon/Makefile clean
-	$(MAKE) PG_CONFIG=$(POSTGRES_INSTALL_DIR)/$*/bin/pg_config \
-	-C $(POSTGRES_INSTALL_DIR)/build/neon-walredo-$* \
-	-f $(ROOT_PROJECT_DIR)/pgxn/neon_walredo/Makefile clean
-	$(MAKE) PG_CONFIG=$(POSTGRES_INSTALL_DIR)/$*/bin/pg_config \
-	-C $(POSTGRES_INSTALL_DIR)/build/neon-test-utils-$* \
-	-f $(ROOT_PROJECT_DIR)/pgxn/neon_test_utils/Makefile clean
-	$(MAKE) PG_CONFIG=$(POSTGRES_INSTALL_DIR)/$*/bin/pg_config \
-	-C $(POSTGRES_INSTALL_DIR)/build/neon-utils-$* \
-	-f $(ROOT_PROJECT_DIR)/pgxn/neon_utils/Makefile clean
+	$(MAKE) -C $(POSTGRES_INSTALL_DIR)/pgxn/neon-$* -f $(ROOT_PROJECT_DIR)/pgxn/neon/Makefile clean
+	$(MAKE) -C $(POSTGRES_INSTALL_DIR)/pgxn/neon_walredo-$* -f $(ROOT_PROJECT_DIR)/pgxn/neon_walredo/Makefile clean
+	$(MAKE) -C $(POSTGRES_INSTALL_DIR)/pgxn/neon_test_utils-$* -f $(ROOT_PROJECT_DIR)/pgxn/neon_test_utils/Makefile clean
 
 .PHONY: neon-pg-ext
 neon-pg-ext: \

README.md

@@ -34,11 +34,6 @@ dnf install flex bison readline-devel zlib-devel openssl-devel \
   libseccomp-devel perl clang cmake postgresql postgresql-contrib protobuf-compiler \
   protobuf-devel
 ```
-* On Arch based systems, these packages are needed:
-```bash
-pacman -S base-devel readline zlib libseccomp openssl clang \
-postgresql-libs cmake postgresql protobuf
-```
 
 2. [Install Rust](https://www.rust-lang.org/tools/install)
 ```
@@ -46,14 +41,11 @@ postgresql-libs cmake postgresql protobuf
 curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
 ```
 
-#### Installing dependencies on macOS (12.3.1)
+#### Installing dependencies on OSX (12.3.1)
 1. Install XCode and dependencies
 ```
 xcode-select --install
 brew install protobuf openssl flex bison
-
-# add openssl to PATH, required for ed25519 keys generation in neon_local
-echo 'export PATH="$(brew --prefix openssl)/bin:$PATH"' >> ~/.zshrc
 ```
 
 2. [Install Rust](https://www.rust-lang.org/tools/install)
@@ -91,10 +83,9 @@ cd neon
 
 # The preferred and default is to make a debug build. This will create a
 # demonstrably slower build than a release build. For a release build,
-# use "BUILD_TYPE=release make -j`nproc` -s"
-# Remove -s for the verbose build log
+# use "BUILD_TYPE=release make -j`nproc`"
 
-make -j`nproc` -s
+make -j`nproc`
 ```
 
 #### Building on OSX
@@ -108,17 +99,16 @@ cd neon
 
 # The preferred and default is to make a debug build. This will create a
 # demonstrably slower build than a release build. For a release build,
-# use "BUILD_TYPE=release make -j`sysctl -n hw.logicalcpu` -s"
-# Remove -s for the verbose build log
+# use "BUILD_TYPE=release make -j`sysctl -n hw.logicalcpu`"
 
-make -j`sysctl -n hw.logicalcpu` -s
+make -j`sysctl -n hw.logicalcpu`
 ```
 
 #### Dependency installation notes
 To run the `psql` client, install the `postgresql-client` package or modify `PATH` and `LD_LIBRARY_PATH` to include `pg_install/bin` and `pg_install/lib`, respectively.
 
 To run the integration tests or Python scripts (not required to use the code), install
-Python (3.9 or higher), and install python3 packages using `./scripts/pysync` (requires [poetry>=1.3](https://python-poetry.org/)) in the project directory.
+Python (3.9 or higher), and install python3 packages using `./scripts/pysync` (requires [poetry](https://python-poetry.org/)) in the project directory.
 
 
 #### Running neon database

compute_tools/Cargo.toml

@@ -11,20 +11,15 @@ clap.workspace = true
 futures.workspace = true
 hyper = { workspace = true, features = ["full"] }
 notify.workspace = true
-num_cpus.workspace = true
-opentelemetry.workspace = true
 postgres.workspace = true
 regex.workspace = true
 serde.workspace = true
 serde_json.workspace = true
 tar.workspace = true
-reqwest = { workspace = true, features = ["json"] }
 tokio = { workspace = true, features = ["rt", "rt-multi-thread"] }
 tokio-postgres.workspace = true
 tracing.workspace = true
-tracing-opentelemetry.workspace = true
 tracing-subscriber.workspace = true
-tracing-utils.workspace = true
 url.workspace = true
 
 workspace_hack.workspace = true

compute_tools/src/bin/compute_ctl.rs

@@ -44,6 +44,7 @@ use tracing::{error, info};
 
 use compute_tools::compute::{ComputeMetrics, ComputeNode, ComputeState, ComputeStatus};
 use compute_tools::http::api::launch_http_server;
+use compute_tools::informant::spawn_vm_informant_if_present;
 use compute_tools::logger::*;
 use compute_tools::monitor::launch_monitor;
 use compute_tools::params::*;
@@ -52,7 +53,7 @@ use compute_tools::spec::*;
 use url::Url;
 
 fn main() -> Result<()> {
-    init_tracing_and_logging(DEFAULT_LOG_LEVEL)?;
+    init_logger(DEFAULT_LOG_LEVEL)?;
 
     let matches = cli().get_matches();
 
@@ -65,9 +66,6 @@ fn main() -> Result<()> {
     let spec = matches.get_one::<String>("spec");
     let spec_path = matches.get_one::<String>("spec-path");
 
-    let compute_id = matches.get_one::<String>("compute-id");
-    let control_plane_uri = matches.get_one::<String>("control-plane-uri");
-
     // Try to use just 'postgres' if no path is provided
     let pgbin = matches.get_one::<String>("pgbin").unwrap();
 
@@ -80,60 +78,17 @@ fn main() -> Result<()> {
                 let path = Path::new(sp);
                 let file = File::open(path)?;
                 serde_json::from_reader(file)?
-            } else if let Some(id) = compute_id {
-                if let Some(cp_base) = control_plane_uri {
-                    let cp_uri = format!("{cp_base}/management/api/v1/{id}/spec");
-                    let jwt: String = match std::env::var("NEON_CONSOLE_JWT") {
-                        Ok(v) => v,
-                        Err(_) => "".to_string(),
-                    };
-
-                    reqwest::blocking::Client::new()
-                        .get(cp_uri)
-                        .header("Authorization", jwt)
-                        .send()?
-                        .json()?
-                } else {
-                    panic!(
-                        "must specify --control-plane-uri \"{:#?}\" and --compute-id \"{:#?}\"",
-                        control_plane_uri, compute_id
-                    );
-                }
             } else {
-                panic!("compute spec should be provided via --spec or --spec-path argument");
+                panic!("cluster spec should be provided via --spec or --spec-path argument");
             }
         }
     };
 
-    // Extract OpenTelemetry context for the startup actions from the spec, and
-    // attach it to the current tracing context.
-    //
-    // This is used to propagate the context for the 'start_compute' operation
-    // from the neon control plane. This allows linking together the wider
-    // 'start_compute' operation that creates the compute container, with the
-    // startup actions here within the container.
-    //
-    // Switch to the startup context here, and exit it once the startup has
-    // completed and Postgres is up and running.
-    //
-    // NOTE: This is supposed to only cover the *startup* actions. Once
-    // postgres is configured and up-and-running, we exit this span. Any other
-    // actions that are performed on incoming HTTP requests, for example, are
-    // performed in separate spans.
-    let startup_context_guard = if let Some(ref carrier) = spec.startup_tracing_context {
-        use opentelemetry::propagation::TextMapPropagator;
-        use opentelemetry::sdk::propagation::TraceContextPropagator;
-        Some(TraceContextPropagator::new().extract(carrier).attach())
-    } else {
-        None
-    };
-
     let pageserver_connstr = spec
         .cluster
         .settings
         .find("neon.pageserver_connstring")
         .expect("pageserver connstr should be provided");
-    let storage_auth_token = spec.storage_auth_token.clone();
     let tenant = spec
         .cluster
         .settings
@@ -154,7 +109,6 @@ fn main() -> Result<()> {
         tenant,
         timeline,
         pageserver_connstr,
-        storage_auth_token,
         metrics: ComputeMetrics::default(),
         state: RwLock::new(ComputeState::new()),
     };
@@ -164,6 +118,8 @@ fn main() -> Result<()> {
     // requests, while configuration is still in progress.
     let _http_handle = launch_http_server(&compute).expect("cannot launch http endpoint thread");
     let _monitor_handle = launch_monitor(&compute).expect("cannot launch compute monitor thread");
+    // Also spawn the thread responsible for handling the VM informant -- if it's present
+    let _vm_informant_handle = spawn_vm_informant_if_present().expect("cannot launch VM informant");
 
     // Start Postgres
     let mut delay_exit = false;
@@ -184,9 +140,6 @@ fn main() -> Result<()> {
     // Wait for the child Postgres process forever. In this state Ctrl+C will
     // propagate to Postgres and it will be shut down as well.
     if let Some(mut pg) = pg {
-        // Startup is finished, exit the startup tracing span
-        drop(startup_context_guard);
-
         let ecode = pg
             .wait()
             .expect("failed to start waiting on Postgres process");
@@ -206,10 +159,6 @@ fn main() -> Result<()> {
         info!("shutting down");
     }
 
-    // Shutdown trace pipeline gracefully, so that it has a chance to send any
-    // pending traces before we exit.
-    tracing_utils::shutdown_tracing();
-
     exit(exit_code.unwrap_or(1))
 }
 
@@ -251,18 +200,6 @@ fn cli() -> clap::Command {
                 .long("spec-path")
                 .value_name("SPEC_PATH"),
         )
-        .arg(
-            Arg::new("compute-id")
-                .short('i')
-                .long("compute-id")
-                .value_name("COMPUTE_ID"),
-        )
-        .arg(
-            Arg::new("control-plane-uri")
-                .short('p')
-                .long("control-plane-uri")
-                .value_name("CONTROL_PLANE"),
-        )
 }
 
 #[test]

compute_tools/src/compute.rs

@@ -18,7 +18,6 @@ use std::fs;
 use std::os::unix::fs::PermissionsExt;
 use std::path::Path;
 use std::process::{Command, Stdio};
-use std::str::FromStr;
 use std::sync::atomic::{AtomicU64, Ordering};
 use std::sync::RwLock;
 
@@ -26,7 +25,6 @@ use anyhow::{Context, Result};
 use chrono::{DateTime, Utc};
 use postgres::{Client, NoTls};
 use serde::{Serialize, Serializer};
-use tokio_postgres;
 use tracing::{info, instrument, warn};
 
 use crate::checker::create_writability_check_data;
@@ -45,7 +43,6 @@ pub struct ComputeNode {
     pub tenant: String,
     pub timeline: String,
     pub pageserver_connstr: String,
-    pub storage_auth_token: Option<String>,
     pub metrics: ComputeMetrics,
     /// Volatile part of the `ComputeNode` so should be used under `RwLock`
     /// to allow HTTP API server to serve status requests, while configuration
@@ -128,18 +125,7 @@ impl ComputeNode {
     fn get_basebackup(&self, lsn: &str) -> Result<()> {
         let start_time = Utc::now();
 
-        let mut config = postgres::Config::from_str(&self.pageserver_connstr)?;
-
-        // Use the storage auth token from the config file, if given.
-        // Note: this overrides any password set in the connection string.
-        if let Some(storage_auth_token) = &self.storage_auth_token {
-            info!("Got storage auth token from spec file");
-            config.password(storage_auth_token);
-        } else {
-            info!("Storage auth token not set");
-        }
-
-        let mut client = config.connect(NoTls)?;
+        let mut client = Client::connect(&self.pageserver_connstr, NoTls)?;
         let basebackup_cmd = match lsn {
             "0/0" => format!("basebackup {} {}", &self.tenant, &self.timeline), // First start of the compute
             _ => format!("basebackup {} {} {}", &self.tenant, &self.timeline, lsn),
@@ -176,11 +162,6 @@ impl ComputeNode {
         let sync_handle = Command::new(&self.pgbin)
             .args(["--sync-safekeepers"])
             .env("PGDATA", &self.pgdata) // we cannot use -D in this mode
-            .envs(if let Some(storage_auth_token) = &self.storage_auth_token {
-                vec![("NEON_AUTH_TOKEN", storage_auth_token)]
-            } else {
-                vec![]
-            })
             .stdout(Stdio::piped())
             .spawn()
             .expect("postgres --sync-safekeepers failed to start");
@@ -258,11 +239,6 @@ impl ComputeNode {
         // Run postgres as a child process.
         let mut pg = Command::new(&self.pgbin)
             .args(["-D", &self.pgdata])
-            .envs(if let Some(storage_auth_token) = &self.storage_auth_token {
-                vec![("NEON_AUTH_TOKEN", storage_auth_token)]
-            } else {
-                vec![]
-            })
             .spawn()
             .expect("cannot start postgres process");
 
@@ -308,7 +284,6 @@ impl ComputeNode {
         handle_role_deletions(self, &mut client)?;
         handle_grants(self, &mut client)?;
         create_writability_check_data(&mut client)?;
-        handle_extensions(&self.spec, &mut client)?;
 
         // 'Close' connection
         drop(client);
@@ -425,43 +400,4 @@ impl ComputeNode {
 
         Ok(())
     }
-
-    /// Select `pg_stat_statements` data and return it as a stringified JSON
-    pub async fn collect_insights(&self) -> String {
-        let mut result_rows: Vec<String> = Vec::new();
-        let connect_result = tokio_postgres::connect(self.connstr.as_str(), NoTls).await;
-        let (client, connection) = connect_result.unwrap();
-        tokio::spawn(async move {
-            if let Err(e) = connection.await {
-                eprintln!("connection error: {}", e);
-            }
-        });
-        let result = client
-            .simple_query(
-                "SELECT
-    row_to_json(pg_stat_statements)
-FROM
-    pg_stat_statements
-WHERE
-    userid != 'cloud_admin'::regrole::oid
-ORDER BY
-    (mean_exec_time + mean_plan_time) DESC
-LIMIT 100",
-            )
-            .await;
-
-        if let Ok(raw_rows) = result {
-            for message in raw_rows.iter() {
-                if let postgres::SimpleQueryMessage::Row(row) = message {
-                    if let Some(json) = row.get(0) {
-                        result_rows.push(json.to_string());
-                    }
-                }
-            }
-
-            format!("{{\"pg_stat_statements\": [{}]}}", result_rows.join(","))
-        } else {
-            "{{\"pg_stat_statements\": []}}".to_string()
-        }
-    }
 }

compute_tools/src/http/api.rs

@@ -3,22 +3,16 @@ use std::net::SocketAddr;
 use std::sync::Arc;
 use std::thread;
 
-use crate::compute::ComputeNode;
 use anyhow::Result;
 use hyper::service::{make_service_fn, service_fn};
 use hyper::{Body, Method, Request, Response, Server, StatusCode};
-use num_cpus;
 use serde_json;
 use tracing::{error, info};
-use tracing_utils::http::OtelName;
+
+use crate::compute::ComputeNode;
 
 // Service function to handle all available routes.
-async fn routes(req: Request<Body>, compute: &Arc<ComputeNode>) -> Response<Body> {
-    //
-    // NOTE: The URI path is currently included in traces. That's OK because
-    // it doesn't contain any variable parts or sensitive information. But
-    // please keep that in mind if you change the routing here.
-    //
+async fn routes(req: Request<Body>, compute: Arc<ComputeNode>) -> Response<Body> {
     match (req.method(), req.uri().path()) {
         // Serialized compute state.
         (&Method::GET, "/status") => {
@@ -34,33 +28,15 @@ async fn routes(req: Request<Body>, compute: &Arc<ComputeNode>) -> Response<Body
             Response::new(Body::from(serde_json::to_string(&compute.metrics).unwrap()))
         }
 
-        // Collect Postgres current usage insights
-        (&Method::GET, "/insights") => {
-            info!("serving /insights GET request");
-            let insights = compute.collect_insights().await;
-            Response::new(Body::from(insights))
-        }
-
         (&Method::POST, "/check_writability") => {
             info!("serving /check_writability POST request");
-            let res = crate::checker::check_writability(compute).await;
+            let res = crate::checker::check_writability(&compute).await;
             match res {
                 Ok(_) => Response::new(Body::from("true")),
                 Err(e) => Response::new(Body::from(e.to_string())),
             }
         }
 
-        (&Method::GET, "/info") => {
-            let num_cpus = num_cpus::get_physical();
-            info!("serving /info GET request. num_cpus: {}", num_cpus);
-            Response::new(Body::from(
-                serde_json::json!({
-                    "num_cpus": num_cpus,
-                })
-                .to_string(),
-            ))
-        }
-
         // Return the `404 Not Found` for any other routes.
         _ => {
             let mut not_found = Response::new(Body::from("404 Not Found"));
@@ -80,19 +56,7 @@ async fn serve(state: Arc<ComputeNode>) {
         async move {
             Ok::<_, Infallible>(service_fn(move |req: Request<Body>| {
                 let state = state.clone();
-                async move {
-                    Ok::<_, Infallible>(
-                        // NOTE: We include the URI path in the string. It
-                        // doesn't contain any variable parts or sensitive
-                        // information in this API.
-                        tracing_utils::http::tracing_handler(
-                            req,
-                            |req| routes(req, &state),
-                            OtelName::UriPath,
-                        )
-                        .await,
-                    )
-                }
+                async move { Ok::<_, Infallible>(routes(req, state).await) }
             }))
         }
     });

compute_tools/src/http/openapi_spec.yaml

@@ -10,12 +10,12 @@ paths:
   /status:
     get:
       tags:
-      - Info
+      - "info"
       summary: Get compute node internal status
       description: ""
       operationId: getComputeStatus
       responses:
-        200:
+        "200":
           description: ComputeState
           content:
             application/json:
@@ -25,58 +25,27 @@ paths:
   /metrics.json:
     get:
       tags:
-      - Info
+      - "info"
       summary: Get compute node startup metrics in JSON format
       description: ""
       operationId: getComputeMetricsJSON
       responses:
-        200:
+        "200":
           description: ComputeMetrics
           content:
             application/json:
               schema:
                 $ref: "#/components/schemas/ComputeMetrics"
 
-  /insights:
-    get:
-      tags:
-      - Info
-      summary: Get current compute insights in JSON format
-      description: |
-        Note, that this doesn't include any historical data
-      operationId: getComputeInsights
-      responses:
-        200:
-          description: Compute insights
-          content:
-            application/json:
-              schema:
-                $ref: "#/components/schemas/ComputeInsights"
-
-  /info:
-    get:
-      tags:
-      - "info"
-      summary: Get info about the compute Pod/VM
-      description: ""
-      operationId: getInfo
-      responses:
-        "200":
-          description: Info
-          content:
-            application/json:
-              schema:
-                $ref: "#/components/schemas/Info"
-
   /check_writability:
     post:
       tags:
-      - Check
+      - "check"
       summary: Check that we can write new data on this compute
       description: ""
       operationId: checkComputeWritability
       responses:
-        200:
+        "200":
           description: Check result
           content:
             text/plain:
@@ -111,15 +80,6 @@ components:
         total_startup_ms:
           type: integer
 
-    Info:
-      type: object
-      description: Information about VM/Pod
-      required:
-        - num_cpus
-      properties:
-        num_cpus:
-          type: integer
-
     ComputeState:
       type: object
       required:
@@ -136,15 +96,6 @@ components:
           type: string
           description: Text of the error during compute startup, if any
 
-    ComputeInsights:
-      type: object
-      properties:
-        pg_stat_statements:
-          description: Contains raw output from pg_stat_statements in JSON format
-          type: array
-          items:
-            type: object
-
     ComputeStatus:
       type: string
       enum:

compute_tools/src/informant.rs

@@ -0,0 +1,50 @@
+use std::path::Path;
+use std::process;
+use std::thread;
+use std::time::Duration;
+use tracing::{info, warn};
+
+use anyhow::{Context, Result};
+
+const VM_INFORMANT_PATH: &str = "/bin/vm-informant";
+const RESTART_INFORMANT_AFTER_MILLIS: u64 = 5000;
+
+/// Launch a thread to start the VM informant if it's present (and restart, on failure)
+pub fn spawn_vm_informant_if_present() -> Result<Option<thread::JoinHandle<()>>> {
+    let exists = Path::new(VM_INFORMANT_PATH)
+        .try_exists()
+        .context("could not check if path exists")?;
+
+    if !exists {
+        return Ok(None);
+    }
+
+    Ok(Some(
+        thread::Builder::new()
+            .name("run-vm-informant".into())
+            .spawn(move || run_informant())?,
+    ))
+}
+
+fn run_informant() -> ! {
+    let restart_wait = Duration::from_millis(RESTART_INFORMANT_AFTER_MILLIS);
+
+    info!("starting VM informant");
+
+    loop {
+        let mut cmd = process::Command::new(VM_INFORMANT_PATH);
+        // Block on subprocess:
+        let result = cmd.status();
+
+        match result {
+            Err(e) => warn!("failed to run VM informant at {VM_INFORMANT_PATH:?}: {e}"),
+            Ok(status) if !status.success() => {
+                warn!("{VM_INFORMANT_PATH} exited with code {status:?}, retrying")
+            }
+            Ok(_) => info!("{VM_INFORMANT_PATH} ended gracefully (unexpectedly). Retrying"),
+        }
+
+        // Wait before retrying
+        thread::sleep(restart_wait);
+    }
+}

compute_tools/src/lib.rs

@@ -8,6 +8,7 @@ pub mod http;
 #[macro_use]
 pub mod logger;
 pub mod compute;
+pub mod informant;
 pub mod monitor;
 pub mod params;
 pub mod pg_helpers;

compute_tools/src/logger.rs

@@ -1,37 +1,21 @@
-use tracing_opentelemetry::OpenTelemetryLayer;
+use anyhow::Result;
 use tracing_subscriber::layer::SubscriberExt;
 use tracing_subscriber::prelude::*;
 
-/// Initialize logging to stderr, and OpenTelemetry tracing and exporter.
-///
-/// Logging is configured using either `default_log_level` or
+/// Initialize `env_logger` using either `default_level` or
 /// `RUST_LOG` environment variable as default log level.
-///
-/// OpenTelemetry is configured with OTLP/HTTP exporter. It picks up
-/// configuration from environment variables. For example, to change the destination,
-/// set `OTEL_EXPORTER_OTLP_ENDPOINT=http://jaeger:4318`. See
-/// `tracing-utils` package description.
-///
-pub fn init_tracing_and_logging(default_log_level: &str) -> anyhow::Result<()> {
-    // Initialize Logging
+pub fn init_logger(default_level: &str) -> Result<()> {
     let env_filter = tracing_subscriber::EnvFilter::try_from_default_env()
-        .unwrap_or_else(|_| tracing_subscriber::EnvFilter::new(default_log_level));
+        .unwrap_or_else(|_| tracing_subscriber::EnvFilter::new(default_level));
 
     let fmt_layer = tracing_subscriber::fmt::layer()
         .with_target(false)
         .with_writer(std::io::stderr);
 
-    // Initialize OpenTelemetry
-    let otlp_layer =
-        tracing_utils::init_tracing_without_runtime("compute_ctl").map(OpenTelemetryLayer::new);
-
-    // Put it all together
     tracing_subscriber::registry()
         .with(env_filter)
-        .with(otlp_layer)
         .with(fmt_layer)
         .init();
-    tracing::info!("logging and tracing started");
 
     Ok(())
 }

compute_tools/src/pg_helpers.rs

@@ -47,23 +47,12 @@ pub struct GenericOption {
 /// declare a `trait` on it.
 pub type GenericOptions = Option<Vec<GenericOption>>;
 
-/// Escape a string for including it in a SQL literal
-fn escape_literal(s: &str) -> String {
-    s.replace('\'', "''").replace('\\', "\\\\")
-}
-
-/// Escape a string so that it can be used in postgresql.conf.
-/// Same as escape_literal, currently.
-fn escape_conf_value(s: &str) -> String {
-    s.replace('\'', "''").replace('\\', "\\\\")
-}
-
 impl GenericOption {
     /// Represent `GenericOption` as SQL statement parameter.
     pub fn to_pg_option(&self) -> String {
         if let Some(val) = &self.value {
             match self.vartype.as_ref() {
-                "string" => format!("{} '{}'", self.name, escape_literal(val)),
+                "string" => format!("{} '{}'", self.name, val),
                 _ => format!("{} {}", self.name, val),
             }
         } else {
@@ -74,8 +63,6 @@ impl GenericOption {
     /// Represent `GenericOption` as configuration option.
     pub fn to_pg_setting(&self) -> String {
         if let Some(val) = &self.value {
-            // TODO: check in the console DB that we don't have these settings
-            // set for any non-deleted project and drop this override.
             let name = match self.name.as_str() {
                 "safekeepers" => "neon.safekeepers",
                 "wal_acceptor_reconnect" => "neon.safekeeper_reconnect_timeout",
@@ -84,7 +71,7 @@ impl GenericOption {
             };
 
             match self.vartype.as_ref() {
-                "string" => format!("{} = '{}'", name, escape_conf_value(val)),
+                "string" => format!("{} = '{}'", name, val),
                 _ => format!("{} = {}", name, val),
             }
         } else {
@@ -120,7 +107,6 @@ impl PgOptionsSerialize for GenericOptions {
                 .map(|op| op.to_pg_setting())
                 .collect::<Vec<String>>()
                 .join("\n")
-                + "\n" // newline after last setting
         } else {
             "".to_string()
         }

compute_tools/src/spec.rs

@@ -1,4 +1,3 @@
-use std::collections::HashMap;
 use std::path::Path;
 use std::str::FromStr;
 
@@ -23,10 +22,6 @@ pub struct ComputeSpec {
     /// Expected cluster state at the end of transition process.
     pub cluster: Cluster,
     pub delta_operations: Option<Vec<DeltaOp>>,
-
-    pub storage_auth_token: Option<String>,
-
-    pub startup_tracing_context: Option<HashMap<String, String>>,
 }
 
 /// Cluster state seen from the perspective of the external tools
@@ -389,13 +384,13 @@ pub fn handle_databases(spec: &ComputeSpec, client: &mut Client) -> Result<()> {
                     name.pg_quote(),
                     db.owner.pg_quote()
                 );
-                let _guard = info_span!("executing", query).entered();
+                let _ = info_span!("executing", query).entered();
                 client.execute(query.as_str(), &[])?;
             }
             DatabaseAction::Create => {
                 let mut query: String = format!("CREATE DATABASE {} ", name.pg_quote());
                 query.push_str(&db.to_pg_options());
-                let _guard = info_span!("executing", query).entered();
+                let _ = info_span!("executing", query).entered();
                 client.execute(query.as_str(), &[])?;
             }
         };
@@ -517,18 +512,3 @@ pub fn handle_grants(node: &ComputeNode, client: &mut Client) -> Result<()> {
 
     Ok(())
 }
-
-/// Create required system extensions
-#[instrument(skip_all)]
-pub fn handle_extensions(spec: &ComputeSpec, client: &mut Client) -> Result<()> {
-    if let Some(libs) = spec.cluster.settings.find("shared_preload_libraries") {
-        if libs.contains("pg_stat_statements") {
-            // Create extension only if this compute really needs it
-            let query = "CREATE EXTENSION IF NOT EXISTS pg_stat_statements";
-            info!("creating system extensions with query: {}", query);
-            client.simple_query(query)?;
-        }
-    }
-
-    Ok(())
-}

compute_tools/tests/cluster_spec.json

@@ -178,11 +178,6 @@
                 "name": "neon.pageserver_connstring",
                 "value": "host=127.0.0.1 port=6400",
                 "vartype": "string"
-            },
-            {
-                "name": "test.escaping",
-                "value": "here's a backslash \\ and a quote ' and a double-quote \" hooray",
-                "vartype": "string"
             }
         ]
     },

compute_tools/tests/pg_helpers_tests.rs

@@ -28,30 +28,7 @@ mod pg_helpers_tests {
 
         assert_eq!(
             spec.cluster.settings.as_pg_settings(),
-            r#"fsync = off
-wal_level = replica
-hot_standby = on
-neon.safekeepers = '127.0.0.1:6502,127.0.0.1:6503,127.0.0.1:6501'
-wal_log_hints = on
-log_connections = on
-shared_buffers = 32768
-port = 55432
-max_connections = 100
-max_wal_senders = 10
-listen_addresses = '0.0.0.0'
-wal_sender_timeout = 0
-password_encryption = md5
-maintenance_work_mem = 65536
-max_parallel_workers = 8
-max_worker_processes = 8
-neon.tenant_id = 'b0554b632bd4d547a63b86c3630317e8'
-max_replication_slots = 10
-neon.timeline_id = '2414a61ffc94e428f14b5758fe308e13'
-shared_preload_libraries = 'neon'
-synchronous_standby_names = 'walproposer'
-neon.pageserver_connstring = 'host=127.0.0.1 port=6400'
-test.escaping = 'here''s a backslash \\ and a quote '' and a double-quote " hooray'
-"#
+            "fsync = off\nwal_level = replica\nhot_standby = on\nneon.safekeepers = '127.0.0.1:6502,127.0.0.1:6503,127.0.0.1:6501'\nwal_log_hints = on\nlog_connections = on\nshared_buffers = 32768\nport = 55432\nmax_connections = 100\nmax_wal_senders = 10\nlisten_addresses = '0.0.0.0'\nwal_sender_timeout = 0\npassword_encryption = md5\nmaintenance_work_mem = 65536\nmax_parallel_workers = 8\nmax_worker_processes = 8\nneon.tenant_id = 'b0554b632bd4d547a63b86c3630317e8'\nmax_replication_slots = 10\nneon.timeline_id = '2414a61ffc94e428f14b5758fe308e13'\nshared_preload_libraries = 'neon'\nsynchronous_standby_names = 'walproposer'\nneon.pageserver_connstring = 'host=127.0.0.1 port=6400'"
         );
     }
 

control_plane/Cargo.toml

@@ -15,7 +15,6 @@ postgres.workspace = true
 regex.workspace = true
 reqwest = { workspace = true, features = ["blocking", "json"] }
 serde.workspace = true
-serde_json.workspace = true
 serde_with.workspace = true
 tar.workspace = true
 thiserror.workspace = true
@@ -24,7 +23,6 @@ url.workspace = true
 # Note: Do not directly depend on pageserver or safekeeper; use pageserver_api or safekeeper_api
 # instead, so that recompile times are better.
 pageserver_api.workspace = true
-postgres_backend.workspace = true
 safekeeper_api.workspace = true
 postgres_connection.workspace = true
 storage_broker.workspace = true

control_plane/safekeepers.conf

@@ -2,8 +2,7 @@
 [pageserver]
 listen_pg_addr = '127.0.0.1:64000'
 listen_http_addr = '127.0.0.1:9898'
-pg_auth_type = 'Trust'
-http_auth_type = 'Trust'
+auth_type = 'Trust'
 
 [[safekeepers]]
 id = 1

control_plane/simple.conf

@@ -3,8 +3,7 @@
 [pageserver]
 listen_pg_addr = '127.0.0.1:64000'
 listen_http_addr = '127.0.0.1:9898'
-pg_auth_type = 'Trust'
-http_auth_type = 'Trust'
+auth_type = 'Trust'
 
 [[safekeepers]]
 id = 1

control_plane/src/bin/neon_local.rs

@@ -17,7 +17,6 @@ use pageserver_api::{
     DEFAULT_HTTP_LISTEN_ADDR as DEFAULT_PAGESERVER_HTTP_ADDR,
     DEFAULT_PG_LISTEN_ADDR as DEFAULT_PAGESERVER_PG_ADDR,
 };
-use postgres_backend::AuthType;
 use safekeeper_api::{
     DEFAULT_HTTP_LISTEN_PORT as DEFAULT_SAFEKEEPER_HTTP_PORT,
     DEFAULT_PG_LISTEN_PORT as DEFAULT_SAFEKEEPER_PG_PORT,
@@ -31,6 +30,7 @@ use utils::{
     auth::{Claims, Scope},
     id::{NodeId, TenantId, TenantTimelineId, TimelineId},
     lsn::Lsn,
+    postgres_backend::AuthType,
     project_git_version,
 };
 
@@ -53,15 +53,14 @@ listen_addr = '{DEFAULT_BROKER_ADDR}'
 id = {DEFAULT_PAGESERVER_ID}
 listen_pg_addr = '{DEFAULT_PAGESERVER_PG_ADDR}'
 listen_http_addr = '{DEFAULT_PAGESERVER_HTTP_ADDR}'
-pg_auth_type = '{trust_auth}'
-http_auth_type = '{trust_auth}'
+auth_type = '{pageserver_auth_type}'
 
 [[safekeepers]]
 id = {DEFAULT_SAFEKEEPER_ID}
 pg_port = {DEFAULT_SAFEKEEPER_PG_PORT}
 http_port = {DEFAULT_SAFEKEEPER_HTTP_PORT}
 "#,
-        trust_auth = AuthType::Trust,
+        pageserver_auth_type = AuthType::Trust,
     )
 }
 
@@ -628,7 +627,7 @@ fn handle_pg(pg_match: &ArgMatches, env: &local_env::LocalEnv) -> Result<()> {
 
             let node = cplane.nodes.get(&(tenant_id, node_name.to_string()));
 
-            let auth_token = if matches!(env.pageserver.pg_auth_type, AuthType::NeonJWT) {
+            let auth_token = if matches!(env.pageserver.auth_type, AuthType::NeonJWT) {
                 let claims = Claims::new(Some(tenant_id), Scope::Tenant);
 
                 Some(env.generate_auth_token(&claims)?)

control_plane/src/compute.rs

@@ -14,6 +14,7 @@ use anyhow::{Context, Result};
 use utils::{
     id::{TenantId, TimelineId},
     lsn::Lsn,
+    postgres_backend::AuthType,
 };
 
 use crate::local_env::{LocalEnv, DEFAULT_PG_VERSION};
@@ -96,7 +97,7 @@ impl ComputeControlPlane {
         });
 
         node.create_pgdata()?;
-        node.setup_pg_conf()?;
+        node.setup_pg_conf(self.env.pageserver.auth_type)?;
 
         self.nodes
             .insert((tenant_id, node.name.clone()), Arc::clone(&node));
@@ -277,7 +278,7 @@ impl PostgresNode {
 
     // Write postgresql.conf with default configuration
     // and PG_VERSION file to the data directory of a new node.
-    fn setup_pg_conf(&self) -> Result<()> {
+    fn setup_pg_conf(&self, auth_type: AuthType) -> Result<()> {
         let mut conf = PostgresConf::new();
         conf.append("max_wal_senders", "10");
         conf.append("wal_log_hints", "off");
@@ -301,12 +302,29 @@ impl PostgresNode {
             let config = &self.pageserver.pg_connection_config;
             let (host, port) = (config.host(), config.port());
 
-            // NOTE: avoid spaces in connection string, because it is less error prone if we forward it somewhere.
-            format!("postgresql://no_user@{host}:{port}")
+            // Set up authentication
+            //
+            // $NEON_AUTH_TOKEN will be replaced with value from environment
+            // variable during compute pg startup. It is done this way because
+            // otherwise user will be able to retrieve the value using SHOW
+            // command or pg_settings
+            let password = if let AuthType::NeonJWT = auth_type {
+                "$NEON_AUTH_TOKEN"
+            } else {
+                ""
+            };
+            // NOTE avoiding spaces in connection string, because it is less error prone if we forward it somewhere.
+            // Also note that not all parameters are supported here. Because in compute we substitute $NEON_AUTH_TOKEN
+            // We parse this string and build it back with token from env var, and for simplicity rebuild
+            // uses only needed variables namely host, port, user, password.
+            format!("postgresql://no_user:{password}@{host}:{port}")
         };
         conf.append("shared_preload_libraries", "neon");
         conf.append_line("");
         conf.append("neon.pageserver_connstring", &pageserver_connstr);
+        if let AuthType::NeonJWT = auth_type {
+            conf.append("neon.safekeeper_token_env", "$NEON_AUTH_TOKEN");
+        }
         conf.append("neon.tenant_id", &self.tenant_id.to_string());
         conf.append("neon.timeline_id", &self.timeline_id.to_string());
         if let Some(lsn) = self.lsn {
@@ -429,8 +447,6 @@ impl PostgresNode {
             "DYLD_LIBRARY_PATH",
             self.env.pg_lib_dir(self.pg_version)?.to_str().unwrap(),
         );
-
-        // Pass authentication token used for the connections to pageserver and safekeepers
         if let Some(token) = auth_token {
             cmd.env("NEON_AUTH_TOKEN", token);
         }

control_plane/src/local_env.rs

@@ -5,7 +5,6 @@
 
 use anyhow::{bail, ensure, Context};
 
-use postgres_backend::AuthType;
 use reqwest::Url;
 use serde::{Deserialize, Serialize};
 use serde_with::{serde_as, DisplayFromStr};
@@ -18,8 +17,9 @@ use std::net::SocketAddr;
 use std::path::{Path, PathBuf};
 use std::process::{Command, Stdio};
 use utils::{
-    auth::{encode_from_key_file, Claims},
+    auth::{encode_from_key_file, Claims, Scope},
     id::{NodeId, TenantId, TenantTimelineId, TimelineId},
+    postgres_backend::AuthType,
 };
 
 use crate::safekeeper::SafekeeperNode;
@@ -110,14 +110,15 @@ impl NeonBroker {
 pub struct PageServerConf {
     // node id
     pub id: NodeId,
-
     // Pageserver connection settings
     pub listen_pg_addr: String,
     pub listen_http_addr: String,
 
-    // auth type used for the PG and HTTP ports
-    pub pg_auth_type: AuthType,
-    pub http_auth_type: AuthType,
+    // used to determine which auth type is used
+    pub auth_type: AuthType,
+
+    // jwt auth token used for communication with pageserver
+    pub auth_token: String,
 }
 
 impl Default for PageServerConf {
@@ -126,8 +127,8 @@ impl Default for PageServerConf {
             id: NodeId(0),
             listen_pg_addr: String::new(),
             listen_http_addr: String::new(),
-            pg_auth_type: AuthType::Trust,
-            http_auth_type: AuthType::Trust,
+            auth_type: AuthType::Trust,
+            auth_token: String::new(),
         }
     }
 }
@@ -400,33 +401,48 @@ impl LocalEnv {
 
         fs::create_dir(base_path)?;
 
-        // Generate keypair for JWT.
-        //
-        // The keypair is only needed if authentication is enabled in any of the
-        // components. For convenience, we generate the keypair even if authentication
-        // is not enabled, so that you can easily enable it after the initialization
-        // step. However, if the key generation fails, we treat it as non-fatal if
-        // authentication was not enabled.
+        // generate keys for jwt
+        // openssl genrsa -out private_key.pem 2048
+        let private_key_path;
         if self.private_key_path == PathBuf::new() {
-            match generate_auth_keys(
-                base_path.join("auth_private_key.pem").as_path(),
-                base_path.join("auth_public_key.pem").as_path(),
-            ) {
-                Ok(()) => {
-                    self.private_key_path = PathBuf::from("auth_private_key.pem");
-                }
-                Err(e) => {
-                    if !self.auth_keys_needed() {
-                        eprintln!("Could not generate keypair for JWT authentication: {e}");
-                        eprintln!("Continuing anyway because authentication was not enabled");
-                        self.private_key_path = PathBuf::from("auth_private_key.pem");
-                    } else {
-                        return Err(e);
-                    }
-                }
+            private_key_path = base_path.join("auth_private_key.pem");
+            let keygen_output = Command::new("openssl")
+                .arg("genrsa")
+                .args(["-out", private_key_path.to_str().unwrap()])
+                .arg("2048")
+                .stdout(Stdio::null())
+                .output()
+                .context("failed to generate auth private key")?;
+            if !keygen_output.status.success() {
+                bail!(
+                    "openssl failed: '{}'",
+                    String::from_utf8_lossy(&keygen_output.stderr)
+                );
+            }
+            self.private_key_path = PathBuf::from("auth_private_key.pem");
+
+            let public_key_path = base_path.join("auth_public_key.pem");
+            // openssl rsa -in private_key.pem -pubout -outform PEM -out public_key.pem
+            let keygen_output = Command::new("openssl")
+                .arg("rsa")
+                .args(["-in", private_key_path.to_str().unwrap()])
+                .arg("-pubout")
+                .args(["-outform", "PEM"])
+                .args(["-out", public_key_path.to_str().unwrap()])
+                .stdout(Stdio::null())
+                .output()
+                .context("failed to generate auth private key")?;
+            if !keygen_output.status.success() {
+                bail!(
+                    "openssl failed: '{}'",
+                    String::from_utf8_lossy(&keygen_output.stderr)
+                );
             }
         }
 
+        self.pageserver.auth_token =
+            self.generate_auth_token(&Claims::new(None, Scope::PageServerApi))?;
+
         fs::create_dir_all(self.pg_data_dirs_path())?;
 
         for safekeeper in &self.safekeepers {
@@ -435,12 +451,6 @@ impl LocalEnv {
 
         self.persist_config(base_path)
     }
-
-    fn auth_keys_needed(&self) -> bool {
-        self.pageserver.pg_auth_type == AuthType::NeonJWT
-            || self.pageserver.http_auth_type == AuthType::NeonJWT
-            || self.safekeepers.iter().any(|sk| sk.auth_enabled)
-    }
 }
 
 fn base_path() -> PathBuf {
@@ -450,43 +460,6 @@ fn base_path() -> PathBuf {
     }
 }
 
-/// Generate a public/private key pair for JWT authentication
-fn generate_auth_keys(private_key_path: &Path, public_key_path: &Path) -> anyhow::Result<()> {
-    // Generate the key pair
-    //
-    // openssl genpkey -algorithm ed25519 -out auth_private_key.pem
-    let keygen_output = Command::new("openssl")
-        .arg("genpkey")
-        .args(["-algorithm", "ed25519"])
-        .args(["-out", private_key_path.to_str().unwrap()])
-        .stdout(Stdio::null())
-        .output()
-        .context("failed to generate auth private key")?;
-    if !keygen_output.status.success() {
-        bail!(
-            "openssl failed: '{}'",
-            String::from_utf8_lossy(&keygen_output.stderr)
-        );
-    }
-    // Extract the public key from the private key file
-    //
-    // openssl pkey -in auth_private_key.pem -pubout -out auth_public_key.pem
-    let keygen_output = Command::new("openssl")
-        .arg("pkey")
-        .args(["-in", private_key_path.to_str().unwrap()])
-        .arg("-pubout")
-        .args(["-out", public_key_path.to_str().unwrap()])
-        .output()
-        .context("failed to extract public key from private key")?;
-    if !keygen_output.status.success() {
-        bail!(
-            "openssl failed: '{}'",
-            String::from_utf8_lossy(&keygen_output.stderr)
-        );
-    }
-    Ok(())
-}
-
 #[cfg(test)]
 mod tests {
     use super::*;

control_plane/src/pageserver.rs

@@ -11,7 +11,6 @@ use anyhow::{bail, Context};
 use pageserver_api::models::{
     TenantConfigRequest, TenantCreateRequest, TenantInfo, TimelineCreateRequest, TimelineInfo,
 };
-use postgres_backend::AuthType;
 use postgres_connection::{parse_host_port, PgConnectionConfig};
 use reqwest::blocking::{Client, RequestBuilder, Response};
 use reqwest::{IntoUrl, Method};
@@ -21,6 +20,7 @@ use utils::{
     http::error::HttpErrorBody,
     id::{TenantId, TimelineId},
     lsn::Lsn,
+    postgres_backend::AuthType,
 };
 
 use crate::{background_process, local_env::LocalEnv};
@@ -82,8 +82,15 @@ impl PageServerNode {
         let (host, port) = parse_host_port(&env.pageserver.listen_pg_addr)
             .expect("Unable to parse listen_pg_addr");
         let port = port.unwrap_or(5432);
+        let password = if env.pageserver.auth_type == AuthType::NeonJWT {
+            Some(env.pageserver.auth_token.clone())
+        } else {
+            None
+        };
+
         Self {
-            pg_connection_config: PgConnectionConfig::new_host_port(host, port),
+            pg_connection_config: PgConnectionConfig::new_host_port(host, port)
+                .set_password(password),
             env: env.clone(),
             http_client: Client::new(),
             http_base_url: format!("http://{}/v1", env.pageserver.listen_http_addr),
@@ -99,32 +106,25 @@ impl PageServerNode {
             self.env.pg_distrib_dir_raw().display()
         );
 
-        let http_auth_type_param =
-            format!("http_auth_type='{}'", self.env.pageserver.http_auth_type);
+        let authg_type_param = format!("auth_type='{}'", self.env.pageserver.auth_type);
         let listen_http_addr_param = format!(
             "listen_http_addr='{}'",
             self.env.pageserver.listen_http_addr
         );
-
-        let pg_auth_type_param = format!("pg_auth_type='{}'", self.env.pageserver.pg_auth_type);
         let listen_pg_addr_param =
             format!("listen_pg_addr='{}'", self.env.pageserver.listen_pg_addr);
-
         let broker_endpoint_param = format!("broker_endpoint='{}'", self.env.broker.client_url());
 
         let mut overrides = vec![
             id,
             pg_distrib_dir_param,
-            http_auth_type_param,
-            pg_auth_type_param,
+            authg_type_param,
             listen_http_addr_param,
             listen_pg_addr_param,
             broker_endpoint_param,
         ];
 
-        if self.env.pageserver.http_auth_type != AuthType::Trust
-            || self.env.pageserver.pg_auth_type != AuthType::Trust
-        {
+        if self.env.pageserver.auth_type != AuthType::Trust {
             overrides.push("auth_validation_public_key_path='auth_public_key.pem'".to_owned());
         }
         overrides
@@ -247,10 +247,7 @@ impl PageServerNode {
     }
 
     fn pageserver_env_variables(&self) -> anyhow::Result<Vec<(String, String)>> {
-        // FIXME: why is this tied to pageserver's auth type? Whether or not the safekeeper
-        // needs a token, and how to generate that token, seems independent to whether
-        // the pageserver requires a token in incoming requests.
-        Ok(if self.env.pageserver.http_auth_type != AuthType::Trust {
+        Ok(if self.env.pageserver.auth_type != AuthType::Trust {
             // Generate a token to connect from the pageserver to a safekeeper
             let token = self
                 .env
@@ -273,30 +270,27 @@ impl PageServerNode {
         background_process::stop_process(immediate, "pageserver", &self.pid_file())
     }
 
-    pub fn page_server_psql_client(&self) -> anyhow::Result<postgres::Client> {
-        let mut config = self.pg_connection_config.clone();
-        if self.env.pageserver.pg_auth_type == AuthType::NeonJWT {
-            let token = self
-                .env
-                .generate_auth_token(&Claims::new(None, Scope::PageServerApi))?;
-            config = config.set_password(Some(token));
-        }
-        Ok(config.connect_no_tls()?)
+    pub fn page_server_psql(&self, sql: &str) -> Vec<postgres::SimpleQueryMessage> {
+        let mut client = self.pg_connection_config.connect_no_tls().unwrap();
+
+        println!("Pageserver query: '{sql}'");
+        client.simple_query(sql).unwrap()
     }
 
-    fn http_request<U: IntoUrl>(&self, method: Method, url: U) -> anyhow::Result<RequestBuilder> {
+    pub fn page_server_psql_client(&self) -> result::Result<postgres::Client, postgres::Error> {
+        self.pg_connection_config.connect_no_tls()
+    }
+
+    fn http_request<U: IntoUrl>(&self, method: Method, url: U) -> RequestBuilder {
         let mut builder = self.http_client.request(method, url);
-        if self.env.pageserver.http_auth_type == AuthType::NeonJWT {
-            let token = self
-                .env
-                .generate_auth_token(&Claims::new(None, Scope::PageServerApi))?;
-            builder = builder.bearer_auth(token)
+        if self.env.pageserver.auth_type == AuthType::NeonJWT {
+            builder = builder.bearer_auth(&self.env.pageserver.auth_token)
         }
-        Ok(builder)
+        builder
     }
 
     pub fn check_status(&self) -> Result<()> {
-        self.http_request(Method::GET, format!("{}/status", self.http_base_url))?
+        self.http_request(Method::GET, format!("{}/status", self.http_base_url))
             .send()?
             .error_from_body()?;
         Ok(())
@@ -304,7 +298,7 @@ impl PageServerNode {
 
     pub fn tenant_list(&self) -> Result<Vec<TenantInfo>> {
         Ok(self
-            .http_request(Method::GET, format!("{}/tenant", self.http_base_url))?
+            .http_request(Method::GET, format!("{}/tenant", self.http_base_url))
             .send()?
             .error_from_body()?
             .json()?)
@@ -358,16 +352,11 @@ impl PageServerNode {
                 .map(|x| x.parse::<bool>())
                 .transpose()
                 .context("Failed to parse 'trace_read_requests' as bool")?,
-            eviction_policy: settings
-                .get("eviction_policy")
-                .map(|x| serde_json::from_str(x))
-                .transpose()
-                .context("Failed to parse 'eviction_policy' json")?,
         };
         if !settings.is_empty() {
             bail!("Unrecognized tenant settings: {settings:?}")
         }
-        self.http_request(Method::POST, format!("{}/tenant", self.http_base_url))?
+        self.http_request(Method::POST, format!("{}/tenant", self.http_base_url))
             .json(&request)
             .send()?
             .error_from_body()?
@@ -384,7 +373,7 @@ impl PageServerNode {
     }
 
     pub fn tenant_config(&self, tenant_id: TenantId, settings: HashMap<&str, &str>) -> Result<()> {
-        self.http_request(Method::PUT, format!("{}/tenant/config", self.http_base_url))?
+        self.http_request(Method::PUT, format!("{}/tenant/config", self.http_base_url))
             .json(&TenantConfigRequest {
                 tenant_id,
                 checkpoint_distance: settings
@@ -430,11 +419,6 @@ impl PageServerNode {
                     .map(|x| x.parse::<bool>())
                     .transpose()
                     .context("Failed to parse 'trace_read_requests' as bool")?,
-                eviction_policy: settings
-                    .get("eviction_policy")
-                    .map(|x| serde_json::from_str(x))
-                    .transpose()
-                    .context("Failed to parse 'eviction_policy' json")?,
             })
             .send()?
             .error_from_body()?;
@@ -447,7 +431,7 @@ impl PageServerNode {
             .http_request(
                 Method::GET,
                 format!("{}/tenant/{}/timeline", self.http_base_url, tenant_id),
-            )?
+            )
             .send()?
             .error_from_body()?
             .json()?;
@@ -466,7 +450,7 @@ impl PageServerNode {
         self.http_request(
             Method::POST,
             format!("{}/tenant/{}/timeline", self.http_base_url, tenant_id),
-        )?
+        )
         .json(&TimelineCreateRequest {
             new_timeline_id,
             ancestor_start_lsn,
@@ -503,7 +487,7 @@ impl PageServerNode {
         pg_wal: Option<(Lsn, PathBuf)>,
         pg_version: u32,
     ) -> anyhow::Result<()> {
-        let mut client = self.page_server_psql_client()?;
+        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

@@ -1,6 +1,7 @@
 use std::io::Write;
 use std::path::PathBuf;
 use std::process::Child;
+use std::sync::Arc;
 use std::{io, result};
 
 use anyhow::Context;
@@ -10,6 +11,7 @@ use reqwest::{IntoUrl, Method};
 use thiserror::Error;
 use utils::{http::error::HttpErrorBody, id::NodeId};
 
+use crate::pageserver::PageServerNode;
 use crate::{
     background_process,
     local_env::{LocalEnv, SafekeeperConf},
@@ -63,10 +65,14 @@ pub struct SafekeeperNode {
     pub env: LocalEnv,
     pub http_client: Client,
     pub http_base_url: String,
+
+    pub pageserver: Arc<PageServerNode>,
 }
 
 impl SafekeeperNode {
     pub fn from_env(env: &LocalEnv, conf: &SafekeeperConf) -> SafekeeperNode {
+        let pageserver = Arc::new(PageServerNode::from_env(env));
+
         SafekeeperNode {
             id: conf.id,
             conf: conf.clone(),
@@ -74,6 +80,7 @@ impl SafekeeperNode {
             env: env.clone(),
             http_client: Client::new(),
             http_base_url: format!("http://127.0.0.1:{}/v1", conf.http_port),
+            pageserver,
         }
     }
 
@@ -108,10 +115,6 @@ impl SafekeeperNode {
         let datadir = self.datadir_path();
 
         let id_string = id.to_string();
-        // TODO: add availability_zone to the config.
-        // Right now we just specify any value here and use it to check metrics in tests.
-        let availability_zone = format!("sk-{}", id_string);
-
         let mut args = vec![
             "-D",
             datadir.to_str().with_context(|| {
@@ -123,8 +126,6 @@ impl SafekeeperNode {
             &listen_pg,
             "--listen-http",
             &listen_http,
-            "--availability-zone",
-            &availability_zone,
         ];
         if !self.conf.sync {
             args.push("--no-sync");

docker-compose/docker-compose.yml

@@ -160,7 +160,6 @@ services:
     build:
       context: ./compute_wrapper/
       args:
-        - REPOSITORY=${REPOSITORY:-neondatabase}
         - COMPUTE_IMAGE=compute-node-v${PG_VERSION:-14}
         - TAG=${TAG:-latest}
         - http_proxy=$http_proxy

docs/authentication.md

@@ -29,54 +29,12 @@ These components should not have access to the private key and may only get toke
 The key pair is generated once for an installation of compute/pageserver/safekeeper, e.g. by `neon_local init`.
 There is currently no way to rotate the key without bringing down all components.
 
-### Best practices
-
-See [RFC 8725: JSON Web Token Best Current Practices](https://www.rfc-editor.org/rfc/rfc8725)
-
-
-### Token format
-
-The JWT tokens in Neon use "EdDSA" as the algorithm (defined in [RFC8037](https://www.rfc-editor.org/rfc/rfc8037)).
-
-Example:
-
-Header:
-
-```
-{
-  "alg": "EdDSA",
-  "typ": "JWT"
-}
-```
-
-Payload:
-
-```
-{
-  "scope": "tenant",  # "tenant", "pageserverapi", or "safekeeperdata"
-  "tenant_id": "5204921ff44f09de8094a1390a6a50f6",
-}
-```
-
-
-Meanings of scope:
-
-"tenant": Provides access to all data for a specific tenant
-
-"pageserverapi": Provides blanket access to all tenants on the pageserver plus pageserver-wide APIs.
-Should only be used e.g. for status check/tenant creation/list.
-
-"safekeeperdata": Provides blanket access to all data on the safekeeper plus safekeeper-wide APIs.
-Should only be used e.g. for status check.
-Currently also used for connection from any pageserver to any safekeeper.
-
-
 ### CLI
 CLI generates a key pair during call to `neon_local init` with the following commands:
 
 ```bash
-openssl genpkey -algorithm ed25519 -out auth_private_key.pem
-openssl pkey -in auth_private_key.pem -pubout -out auth_public_key.pem
+openssl genrsa -out auth_private_key.pem 2048
+openssl rsa -in auth_private_key.pem -pubout -outform PEM -out auth_public_key.pem
 ```
 
 Configuration files for all components point to `public_key.pem` for JWT validation.
@@ -106,22 +64,20 @@ Their authentication is just plain PostgreSQL authentication and out of scope fo
 There is no administrative API except those provided by PostgreSQL.
 
 #### Outgoing connections
-Compute connects to Pageserver for getting pages. The connection string is
-configured by the `neon.pageserver_connstring` PostgreSQL GUC,
-e.g. `postgresql://no_user@localhost:15028`. If the `$NEON_AUTH_TOKEN`
-environment variable is set, it is used as the password for the connection. (The
-pageserver uses JWT tokens for authentication, so the password is really a
-token.)
+Compute connects to Pageserver for getting pages.
+The connection string is configured by the `neon.pageserver_connstring` PostgreSQL GUC, e.g. `postgresql://no_user:$NEON_AUTH_TOKEN@localhost:15028`.
+The environment variable inside the connection string is substituted with
+the JWT token.
 
-Compute connects to Safekeepers to write and commit data. The list of safekeeper
-addresses is given in the `neon.safekeepers` GUC. The connections to the
-safekeepers take the password from the `$NEON_AUTH_TOKEN` environment
-variable, if set.
+Compute connects to Safekeepers to write and commit data.
+The token is the same for all safekeepers.
+It's stored in an environment variable, whose name is configured
+by the `neon.safekeeper_token_env` PostgreSQL GUC.
+If the GUC is unset, no token is passed.
 
-The `compute_ctl` binary that runs before the PostgreSQL server, and launches
-PostgreSQL, also makes a connection to the pageserver. It uses it to fetch the
-initial "base backup" dump, to initialize the PostgreSQL data directory. It also
-uses `$NEON_AUTH_TOKEN` as the password for the connection.
+Note that both tokens can be (and typically are) the same;
+the scope is the tenant and the token is usually passed through the
+`$NEON_AUTH_TOKEN` environment variable.
 
 ### Pageserver
 #### Overview
@@ -146,12 +102,10 @@ Each compute should present a token valid for the timeline's tenant.
 Pageserver also has HTTP API: some parts are per-tenant,
 some parts are server-wide, these are different scopes.
 
-Authentication can be enabled separately for the HTTP mgmt API, and
-for the libpq connections from compute. The `http_auth_type` and
-`pg_auth_type` configuration variables in Pageserver's config may
-have one of these values:
+The `auth_type` configuration variable in Pageserver's config may have
+either of three values:
 
-* `Trust` removes all authentication.
+* `Trust` removes all authentication. The outdated `MD5` value does likewise
 * `NeonJWT` enables JWT validation.
    Tokens are validated using the public key which lies in a PEM file
    specified in the `auth_validation_public_key_path` config.

docs/docker.md

@@ -37,9 +37,9 @@ 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/
+$ cd docker-compose/docker-compose.yml
 $ docker-compose down   # remove the conainers if exists
-$ PG_VERSION=15 TAG=2937 docker-compose up --build -d  # You can specify the postgres and image version
+$ 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 docker-compose_storage_broker_1       ... done
 (...omit...)

docs/rfcs/022-pageserver-delete-from-s3.md

@@ -1,269 +0,0 @@
-# Deleting pageserver part of tenants data from s3
-
-Created on 08.03.23
-
-## Motivation
-
-Currently we dont delete pageserver part of the data from s3 when project is deleted. (The same is true for safekeepers, but this outside of the scope of this RFC).
-
-This RFC aims to spin a discussion to come to a robust deletion solution that wont put us in into a corner for features like postponed deletion (when we keep data for user to be able to restore a project if it was deleted by accident)
-
-## Summary
-
-TLDR; There are two options, one based on control plane issuing actual delete requests to s3 and the other one that keeps s3 stuff bound to pageserver. Each one has its pros and cons.
-
-The decision is to stick with pageserver centric approach. For motivation see [Decision](#decision).
-
-## Components
-
-pageserver, control-plane
-
-## Requirements
-
-Deletion should successfully finish (eventually) without leaving dangling files in presense of:
-
-- component restarts
-- component outage
-- pageserver loss
-
-## Proposed implementation
-
-Before the options are discussed, note that deletion can be quite long process. For deletion from s3 the obvious choice is [DeleteObjects](https://docs.aws.amazon.com/AmazonS3/latest/API/API_DeleteObjects.html) API call. It allows to batch deletion of up to 1k objects in one API call. So deletion operation linearly depends on number of layer files.
-
-Another design limitation is that there is no cheap `mv` operation available for s3. `mv` from `aws s3 mv` uses `copy(src, dst) + delete(src)`. So `mv`-like operation is not feasible as a building block because it actually amplifies the problem with both duration and resulting cost of the operation.
-
-The case when there are multiple pageservers handling the same tenants is largely out of scope of the RFC. We still consider case with migration from one PS to another, but do not consider case when tenant exists on multiple pageservers for extended period of time. The case with multiple pageservers can be reduced to case with one pageservers by calling detach on all pageservers except the last one, for it actual delete needs to be called.
-
-For simplicity lets look into deleting tenants. Differences in deletion process between tenants and timelines are mentioned in paragraph ["Differences between tenants and timelines"](#differences-between-tenants-and-timelines)
-
-### 1. Pageserver owns deletion machinery
-
-#### The sequence
-
-TLDR; With this approach control plane needs to call delete on a tenant and poll for progress. As much as possible is handled on pageserver. Lets see the sequence.
-
-Happy path:
-
-```mermaid
-sequenceDiagram
-    autonumber
-    participant CP as Control Plane
-    participant PS as Pageserver
-    participant S3
-
-    CP->>PS: Delete tenant
-    PS->>S3: Create deleted mark file at <br> /tenant/meta/deleted
-    PS->>PS: Create deleted mark file locally
-    PS->>CP: Accepted
-    PS->>PS: delete local files other than deleted mark
-    loop Delete layers for each timeline
-        PS->>S3: delete(..)
-        CP->>PS: Finished?
-        PS->>CP: False
-    end
-    PS->>S3: Delete mark file
-    PS->>PS: Delete local mark file
-
-    loop Poll for status
-        CP->>PS: Finished?
-        PS->>CP: True or False
-    end
-```
-
-Why two mark files?
-Remote one is needed for cases when pageserver is lost during deletion so other pageserver can learn the deletion from s3 during attach.
-
-Why local mark file is needed?
-
-If we dont have one, we have two choices, delete local data before deleting the remote part or do that after.
-
-If we delete local data before remote then during restart pageserver wont pick up remote tenant at all because nothing is available locally (pageserver looks for remote conuterparts of locally available tenants).
-
-If we delete local data after remote then at the end of the sequence when remote mark file is deleted if pageserver restart happens then the state is the same to situation when pageserver just missing data on remote without knowing the fact that this data is intended to be deleted. In this case the current behavior is upload everything local-only to remote.
-
-Thus we need local record of tenant being deleted as well.
-
-##### Handle pageserver crashes
-
-Lets explore sequences with various crash points.
-
-Pageserver crashes before `deleted` mark file is persisted in s3:
-
-```mermaid
-sequenceDiagram
-    autonumber
-    participant CP as Control Plane
-    participant PS as Pageserver
-    participant S3
-
-    CP->>PS: Delete tenant
-    note over PS: Crash point 1.
-    CP->>PS: Retry delete request
-
-    PS->>S3: Create deleted mark file at <br> /tenant/meta/deleted
-    PS->>PS: Create deleted mark file locally
-
-    PS->>CP: Accepted
-
-    PS->>PS: delete local files other than deleted mark
-
-    loop Delete layers for each timeline
-        PS->>S3: delete(..)
-        CP->>PS: Finished?
-        PS->>CP: False
-    end
-    PS->>S3: Delete mark file
-    PS->>PS: Delete local mark file
-
-    CP->>PS: Finished?
-    PS->>CP: True
-```
-
-Pageserver crashed when deleted mark was about to be persisted in s3, before Control Plane gets a response:
-
-```mermaid
-sequenceDiagram
-    autonumber
-    participant CP as Control Plane
-    participant PS as Pageserver
-    participant S3
-
-    CP->>PS: Delete tenant
-    PS->>S3: Create deleted mark file at <br> /tenant/meta/deleted
-
-    note over PS: Crash point 2.
-    note over PS: During startup we reconcile <br> with remote and see <br> whether the remote mark exists
-    alt Remote mark exists
-        PS->>PS: create local mark if its missing
-        PS->>PS: delete local files other than deleted mark
-        loop Delete layers for each timeline
-            PS->>S3: delete(..)
-        end
-
-        note over CP: Eventually console should <br> retry delete request
-
-        CP->>PS: Retry delete tenant
-        PS->>CP: Not modified
-    else Mark is missing
-        note over PS: Continue to operate the tenant as if deletion didnt happen
-
-        note over CP: Eventually console should <br> retry delete request
-
-        CP->>PS: Retry delete tenant
-        PS->>S3: Create deleted mark file at <br> /tenant/meta/deleted
-        PS->>CP: Delete tenant
-    end
-
-    PS->>PS: Continue with layer file deletions
-    loop Delete layers for each timeline
-        PS->>S3: delete(..)
-        CP->>PS: Finished?
-        PS->>CP: False
-    end
-
-    PS->>S3: Delete mark file
-    PS->>PS: Delete local mark file
-
-    CP->>PS: Finished?
-    PS->>CP: True
-```
-
-Similar sequence applies when both local and remote marks were persisted but Control Plane still didnt receive a response.
-
-If pageserver crashes after both mark files were deleted then it will reply to control plane status poll request with 404 which should be treated by control plane as success.
-
-The same applies if pageserver crashes in the end, when remote mark is deleted but before local one gets deleted. In this case on restart pageserver moves forward with deletion of local mark and Control Plane will receive 404.
-
-##### Differences between tenants and timelines
-
-For timeline the sequence is the same with the following differences:
-
-- remote delete mark file can be replaced with a boolean "deleted" flag in index_part.json
-- local deletion mark is not needed, because whole tenant is kept locally so situation described in motivation for local mark is impossible
-
-##### Handle pageserver loss
-
-If pageseserver is lost then the deleted tenant should be attached to different pageserver and delete request needs to be retried against new pageserver. Then attach logic is shared with one described for pageserver restarts (local deletion mark wont be available so needs to be created).
-
-##### Restrictions for tenant that is in progress of being deleted
-
-I propose to add another state to tenant/timeline - PendingDelete. This state shouldnt allow executing any operations aside from polling the deletion status.
-
-#### Summary
-
-Pros:
-
-- Storage is not dependent on control plane. Storage can be restarted even if control plane is not working.
-- Allows for easier dogfooding, console can use Neon backed database as primary operational data store. If storage depends on control plane and control plane depends on storage we're stuck.
-- No need to share inner s3 workings with control plane. Pageserver presents api contract and S3 paths are not part of this contract.
-- No need to pass list of alive timelines to attach call. This will be solved by pageserver observing deleted flag. See
-
-Cons:
-
-- Logic is a tricky, needs good testing
-- Anything else?
-
-### 2. Control plane owns deletion machinery
-
-In this case the only action performed on pageserver is removal of local files.
-
-Everything else is done by control plane. The steps are as follows:
-
-1. Control plane marks tenant as "delete pending" in its database
-2. It lists the s3 for all the files and repeatedly calls delete until nothing is left behind
-3. When no files are left marks deletion as completed
-
-In case of restart it selects all tenants marked as "delete pending" and continues the deletion.
-
-For tenants it is simple. For timelines there are caveats.
-
-Assume that the same workflow is used for timelines.
-
-If a tenant gets relocated during timeline deletion the attach call with its current logic will pick up deleted timeline in its half deleted state.
-
-Available options:
-
-- require list of alive timelines to be passed to attach call
-- use the same schema with flag in index_part.json (again part of the caveats around pageserver restart applies). In this case nothing stops pageserver from implementing deletion inside if we already have these deletion marks.
-
-With first option the following problem becomes apparent:
-
-Who is the source of truth regarding timeline liveness?
-
-Imagine:
-PS1 fails.
-PS2 gets assigned the tenant.
-New branch gets created
-PS1 starts up (is it possible or we just recycle it?)
-PS1 is unaware of the new branch. It can either fall back to s3 ls, or ask control plane.
-
-So here comes the dependency of storage on control plane. During restart storage needs to know which timelines are valid for operation. If there is nothing on s3 that can answer that question storage neeeds to ask control plane.
-
-### Summary
-
-Cons:
-
-- Potential thundering herd-like problem during storage restart (requests to control plane)
-- Potential increase in storage startup time (additional request to control plane)
-- Storage startup starts to depend on console
-- Erroneous attach call can attach tenant in half deleted state
-
-Pros:
-
-- Easier to reason about if you dont have to account for pageserver restarts
-
-### Extra notes
-
-There was a concern that having deletion code in pageserver is a littlebit scary, but we need to have this code somewhere. So to me it is equally scary to have that in whatever place it ends up at.
-
-Delayed deletion can be done with both approaches. As discussed with Anna (@stepashka) this is only relevant for tenants (projects) not for timelines. For first approach detach can be called immediately and deletion can be done later with attach + delete. With second approach control plane needs to start the deletion whenever necessary.
-
-## Decision
-
-After discussion in comments I see that we settled on two options (though a bit different from ones described in rfc). First one is the same - pageserver owns as much as possible. The second option is that pageserver owns markers thing, but actual deletion happens in control plane by repeatedly calling ls + delete.
-
-To my mind the only benefit of the latter approach is possible code reuse between safekeepers and pageservers. Otherwise poking around integrating s3 library into control plane, configuring shared knowledge abouth paths in s3 - are the downsides. Another downside of relying on control plane is the testing process. Control plane resides in different repository so it is quite hard to test pageserver related changes there. e2e test suite there doesnt support shutting down pageservers, which are separate docker containers there instead of just processes.
-
-With pageserver owning everything we still give the retry logic to control plane but its easier to duplicate if needed compared to sharing inner s3 workings. We will have needed tests for retry logic in neon repo.
-
-So the decision is to proceed with pageserver centric approach.

docs/settings.md

@@ -16,7 +16,7 @@ listen_http_addr = '127.0.0.1:9898'
 checkpoint_distance = '268435456' # in bytes
 checkpoint_timeout = '10m'
 
-gc_period = '1 hour'
+gc_period = '100 s'
 gc_horizon = '67108864'
 
 max_file_descriptors = '100'
@@ -101,7 +101,7 @@ away.
 
 #### gc_period
 
-Interval at which garbage collection is triggered. Default is 1 hour.
+Interval at which garbage collection is triggered. Default is 100 s.
 
 #### image_creation_threshold
 
@@ -109,7 +109,7 @@ L0 delta layer threshold for L1 image layer creation. Default is 3.
 
 #### pitr_interval
 
-WAL retention duration for PITR branching. Default is 7 days.
+WAL retention duration for PITR branching. Default is 30 days.
 
 #### walreceiver_connect_timeout
 

docs/sourcetree.md

@@ -129,12 +129,13 @@ Run `poetry shell` to activate the virtual environment.
 Alternatively, use `poetry run` to run a single command in the venv, e.g. `poetry run pytest`.
 
 ### Obligatory checks
-We force code formatting via `black`, `ruff`, and type hints via `mypy`.
+We force code formatting via `black`, `isort` and type hints via `mypy`.
 Run the following commands in the repository's root (next to `pyproject.toml`):
 
 ```bash
+poetry run isort .  # Imports are reformatted
 poetry run black .  # All code is reformatted
-poetry run ruff .  # Python linter
+poetry run flake8 .  # Python linter
 poetry run mypy .  # Ensure there are no typing errors
 ```
 

docs/synthetic-size.md

@@ -1,335 +0,0 @@
-# Synthetic size
-
-Neon storage has copy-on-write branching, which makes it difficult to
-answer the question "how large is my database"? To give one reasonable
-answer, we calculate _synthetic size_ for a project.
-
-The calculation is called "synthetic", because it is based purely on
-the user-visible logical size, which is the size that you would see on
-a standalone PostgreSQL installation, and the amount of WAL, which is
-also the same as what you'd see on a standalone PostgreSQL, for the
-same set of updates.
-
-The synthetic size does *not* depend on the actual physical size
-consumed in the storage, or implementation details of the Neon storage
-like garbage collection, compaction and compression.  There is a
-strong *correlation* between the physical size and the synthetic size,
-but the synthetic size is designed to be independent of the
-implementation details, so that any improvements we make in the
-storage system simply reduce our COGS. And vice versa: any bugs or bad
-implementation where we keep more data than we would need to, do not
-change the synthetic size or incur any costs to the user.
-
-The synthetic size is calculated for the whole project. It is not
-straighforward to attribute size to individual branches. See "What is
-the size of an individual branch?" for discussion on those
-difficulties.
-
-The synthetic size is designed to:
-
-- Take into account the copy-on-write nature of the storage. For
-  example, if you create a branch, it doesn't immediately add anything
-  to the synthetic size. It starts to affect the synthetic size only
-  as it diverges from the parent branch.
-
-- Be independent of any implementation details of the storage, like
-  garbage collection, remote storage, or compression.
-
-## Terms & assumptions
-
-- logical size is the size of a branch *at a given point in
-  time*. It's the total size of all tables in all databases, as you
-  see with "\l+" in psql for example, plus the Postgres SLRUs and some
-  small amount of metadata. NOTE that currently, Neon does not include
-  the SLRUs and metadata in the logical size. See comment to `get_current_logical_size_non_incremental()`.
-
-- a "point in time" is defined as an LSN value. You can convert a
-  timestamp to an LSN, but the storage internally works with LSNs.
-
-- PITR horizon can be set per-branch.
-
-- PITR horizon can be set as a time interval, e.g. 5 days or hours, or
-  as amount of WAL, in bytes.  If it's given as a time interval, it's
-  converted to an LSN for the calculation.
-
-- PITR horizon can be set to 0, if you don't want to retain any history.
-
-## Calculation
-
-Inputs to the calculation are:
-- logical size of the database at different points in time,
-- amount of WAL generated, and
-- the PITR horizon settings
-
-The synthetic size is based on an idealistic model of the storage
-system, where we pretend that the storage consists of two things:
-- snapshots, containing a full snapshot of the database, at a given
-  point in time, and
-- WAL.
-
-In the simple case that the project contains just one branch (main),
-and a fixed PITR horizon, the synthetic size is the sum of:
-
-- the logical size of the branch *at the beginning of the PITR
-  horizon*, i.e. at the oldest point that you can still recover to, and
-- the size of the WAL covering the PITR horizon.
-
-The snapshot allows you to recover to the beginning of the PITR
-horizon, and the WAL allows you to recover from that point to any
-point within the horizon.
-
-```
-                             WAL
-   -----------------------#########>
-                          ^
-                       snapshot
-
-Legend:
-  ##### PITR horizon. This is the region that you can still access
-        with Point-in-time query and you can still create branches
-        from.
-  ----- history that has fallen out of the PITR horizon, and can no
-        longer be accessed
-```
-
-NOTE: This is not how the storage system actually works! The actual
-implementation is also based on snapshots and WAL, but the snapshots
-are taken for individual database pages and ranges of pages rather
-than the whole database, and it is much more complicated. This model
-is a reasonable approximation, however, to make the synthetic size a
-useful proxy for the actual storage consumption.
-
-
-## Example: Data is INSERTed
-
-For example, let's assume that your database contained 10 GB of data
-at the beginning of the PITR horizon, and you have since then inserted
-5 GB of additional data into it. The additional insertions of 5 GB of
-data consume roughly 5 GB of WAL. In that case, the synthetic size is:
-
-> 10 GB (snapshot) +  5 GB (WAL) = 15 GB
-
-If you now set the PITR horizon on the project to 0, so that no
-historical data is retained, then the beginning PITR horizon would be
-at the end of the branch, so the size of the snapshot would be
-calculated at the end of the branch, after the insertions. Then the
-synthetic size is:
-
-> 15 GB (snapshot) + 0 GB (WAL) = 15 GB.
-
-In this case, the synthetic size is the same, regardless of the PITR horizon,
-because all the history consists of inserts. The newly inserted data takes
-up the same amount of space, whether it's stored as part of the logical
-snapshot, or as WAL. (*)
-
-(*) This is a rough approximation. In reality, the WAL contains
-headers and other overhead, and on the other hand, the logical
-snapshot includes empty space on pages, so the size of insertions in
-WAL can be smaller or greater than the size of the final table after
-the insertions. But in most cases, it's in the same ballpark.
-
-## Example: Data is DELETEd
-
-Let's look at another example:
-
-Let's start again with a database that contains 10 GB of data. Then,
-you DELETE 5 GB of the data, and run VACUUM to free up the space, so
-that the logical size of the database is now only 5 GB.
-
-Let's assume that the WAL for the deletions and the vacuum take up
-100 MB of space. In that case, the synthetic size of the project is:
-
-> 10 GB (snapshot) + 100 MB (WAL) = 10.1 GB
-
-This is much larger than the logical size of the database after the
-deletions (5 GB). That's because the system still needs to retain the
-deleted data, because it's still accessible to queries and branching
-in the PITR window.
-
-If you now set the PITR horizon to 0 or just wait for time to pass so
-that the data falls out of the PITR horizon, making the deleted data
-inaccessible, the synthetic size shrinks:
-
-> 5 GB (snapshot) + 0 GB (WAL) = 5 GB
-
-
-# Branching
-
-Things get more complicated with branching. Branches in Neon are
-copy-on-write, which is also reflected in the synthetic size.
-
-When you create a branch, it doesn't immediately change the synthetic
-size at all. The branch point is within the PITR horizon, and all the
-data needed to recover to that point in time needs to be retained
-anyway.
-
-However, if you make modifications on the branch, the system needs to
-keep the WAL of those modifications. The WAL is included in the
-synthetic size.
-
-## Example: branch and INSERT
-
-Let's assume that you again start with a 10 GB database.
-On the main branch, you insert 2 GB of data. Then you create
-a branch at that point, and insert another 3 GB of data on the
-main branch, and 1 GB of data on the child branch
-
-```
-  child                 +#####>
-                        |
-                        |    WAL
-  main    ---------###############>
-                   ^
-                snapshot
-```
-
-In this case, the synthetic size consists of:
-- the snapshot at the beginning of the PITR horizon (10 GB)
-- the WAL on the main branch (2 GB + 3 GB = 5 GB)
-- the WAL on the child branch (1 GB)
-
-Total: 16 GB
-
-# Diverging branches
-
-If there is only a small amount of changes in the database on the
-different branches, as in the previous example, the synthetic size
-consists of a snapshot before the branch point, containing all the
-shared data, and the WAL on both branches. However, if the branches
-diverge a lot, it is more efficient to store a separate snapshot of
-branches.
-
-## Example: diverging branches
-
-You start with a 10 GB database. You insert 5 GB of data on the main
-branch. Then you create a branch, and immediately delete all the data
-on the child branch and insert 5 GB of new data to it. Then you do the
-same on the main branch. Let's assume
-that the PITR horizon requires keeping the last 1 GB of WAL on the
-both branches.
-
-```
-                              snapshot
-                                  v     WAL
-  child                 +---------##############>
-                        |
-                        |
-  main     -------------+---------##############>
-                                  ^     WAL
-                              snapshot
-```
-
-In this case, the synthetic size consists of:
-- snapshot at the beginning of the PITR horizon on the main branch (4 GB)
-- WAL on the main branch (1 GB)
-- snapshot at the beginning of the PITR horizon on the child branch (4 GB)
-- last 1 GB of WAL on the child branch (1 GB)
-
-Total: 10 GB
-
-The alternative way to store this would be to take only one snapshot
-at the beginning of branch point, and keep all the WAL on both
-branches.  However, the size with that method would be larger, as it
-would require one 10 GB snapshot, and 5 GB + 5 GB of WAL. It depends
-on the amount of changes (WAL) on both branches, and the logical size
-at the branch point, which method would result in a smaller synthetic
-size. On each branch point, the system performs the calculation with
-both methods, and uses the method that is cheaper, i.e. the one that
-results in a smaller synthetic size.
-
-One way to think about this is that when you create a branch, it
-starts out as a thin branch that only stores the WAL since the branch
-point.  As you modify it, and the amount of WAL grows, at some point
-it becomes cheaper to store a completely new snapshot of the branch
-and truncate the WAL.
-
-
-# What is the size of an individual branch?
-
-Synthetic size is calculated for the whole project, and includes all
-branches. There is no such thing as the size of a branch, because it
-is not straighforward to attribute the parts of size to individual
-branches.
-
-## Example: attributing size to branches
-
-(copied from https://github.com/neondatabase/neon/pull/2884#discussion_r1029365278)
-
-Imagine that you create two branches, A and B, at the same point from
-main branch, and do a couple of small updates on both branches. Then
-six months pass, and during those six months the data on the main
-branch churns over completely multiple times. The retention period is,
-say 1 month.
-
-```
-                      +------> A
-                     /
---------------------*-------------------------------> main
-                     \
-                      +--------> B
-```
-
-In that situation, the synthetic tenant size would be calculated based
-on a "logical snapshot" at the branch point, that is, the logical size
-of the database at that point. Plus the WAL on branches A and B. Let's
-say that the snapshot size is 10 GB, and the WAL is 1 MB on both
-branches A and B. So the total synthetic storage size is 10002
-MB. (Let's ignore the main branch for now, that would be just added to
-the sum)
-
-How would you break that down per branch? I can think of three
-different ways to do it, and all of them have their own problems:
-
-### Subtraction method
-
-For each branch, calculate how much smaller the total synthetic size
-would be, if that branch didn't exist. In other words, how much would
-you save if you dropped the branch. With this method, the size of
-branches A and B is 1 MB.
-
-With this method, the 10 GB shared logical snapshot is not included
-for A nor B. So the size of all branches is not equal to the total
-synthetic size of the tenant. If you drop branch A, you save 1 MB as
-you'd expect, but also the size of B suddenly jumps from 1 MB to 10001
-MB, which might feel surprising.
-
-### Division method
-
-Divide the common parts evenly across all branches that need
-them. With this method, the size of branches A and B would be 5001 MB.
-
-With this method, the sum of all branches adds up to the total
-synthetic size. But it's surprising in other ways: if you drop branch
-A, you might think that you save 5001 MB, but in reality you only save
-1 MB, and the size of branch B suddenly grows from 5001 to 10001 MB.
-
-### Addition method
-
-For each branch, include all the snapshots and WAL that it depends on,
-even if some of them are shared by other branches. With this method,
-the size of branches A and B would be 10001 MB.
-
-The surprise with this method is that the sum of all the branches is
-larger than the total synthetic size. And if you drop branch A, the
-total synthetic size doesn't fall by 10001 MB as you might think.
-
-# Alternatives
-
-A sort of cop-out method would be to show the whole tree of branches
-graphically, and for each section of WAL or logical snapshot, display
-the size of that section. You can then see which branches depend on
-which sections, which sections are shared etc. That would be good to
-have in the UI anyway.
-
-Or perhaps calculate per-branch numbers using the subtraction method,
-and in addition to that, one more number for "shared size" that
-includes all the data that is needed by more than one branch.
-
-## Which is the right method?
-
-The bottom line is that it's not straightforward to attribute the
-synthetic size to individual branches. There are things we can do, and
-all of those methods are pretty straightforward to implement, but they
-all have their own problems. What makes sense depends a lot on what
-you want to do with the number, what question you are trying to
-answer.

libs/metrics/Cargo.toml

@@ -8,6 +8,5 @@ license.workspace = true
 prometheus.workspace = true
 libc.workspace = true
 once_cell.workspace = true
-chrono.workspace = true
 
 workspace_hack.workspace = true

libs/metrics/src/launch_timestamp.rs

@@ -1,34 +0,0 @@
-//! A timestamp captured at process startup to identify restarts of the process, e.g., in logs and metrics.
-
-use chrono::Utc;
-
-use super::register_uint_gauge;
-use std::fmt::Display;
-
-pub struct LaunchTimestamp(chrono::DateTime<Utc>);
-
-impl LaunchTimestamp {
-    pub fn generate() -> Self {
-        LaunchTimestamp(Utc::now())
-    }
-}
-
-impl Display for LaunchTimestamp {
-    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
-        write!(f, "{}", self.0)
-    }
-}
-
-pub fn set_launch_timestamp_metric(launch_ts: &'static LaunchTimestamp) {
-    let millis_since_epoch: u64 = launch_ts
-        .0
-        .timestamp_millis()
-        .try_into()
-        .expect("we're after the epoch, this should be positive");
-    let metric = register_uint_gauge!(
-        "libmetrics_launch_timestamp",
-        "Timestamp (millis since epoch) at wich the process launched."
-    )
-    .unwrap();
-    metric.set(millis_since_epoch);
-}

libs/metrics/src/lib.rs

@@ -8,7 +8,6 @@ pub use prometheus::opts;
 pub use prometheus::register;
 pub use prometheus::{core, default_registry, proto};
 pub use prometheus::{exponential_buckets, linear_buckets};
-pub use prometheus::{register_counter_vec, Counter, CounterVec};
 pub use prometheus::{register_gauge, Gauge};
 pub use prometheus::{register_gauge_vec, GaugeVec};
 pub use prometheus::{register_histogram, Histogram};
@@ -20,7 +19,6 @@ pub use prometheus::{register_int_gauge_vec, IntGaugeVec};
 pub use prometheus::{Encoder, TextEncoder};
 use prometheus::{Registry, Result};
 
-pub mod launch_timestamp;
 mod wrappers;
 pub use wrappers::{CountedReader, CountedWriter};
 
@@ -35,14 +33,6 @@ macro_rules! register_uint_gauge_vec {
     }};
 }
 
-#[macro_export]
-macro_rules! register_uint_gauge {
-    ($NAME:expr, $HELP:expr $(,)?) => {{
-        let gauge = $crate::UIntGauge::new($NAME, $HELP).unwrap();
-        $crate::register(Box::new(gauge.clone())).map(|_| gauge)
-    }};
-}
-
 /// Special internal registry, to collect metrics independently from the default registry.
 /// Was introduced to fix deadlock with lazy registration of metrics in the default registry.
 static INTERNAL_REGISTRY: Lazy<Registry> = Lazy::new(Registry::new);

libs/pageserver_api/Cargo.toml

@@ -13,7 +13,5 @@ bytes.workspace = true
 byteorder.workspace = true
 utils.workspace = true
 postgres_ffi.workspace = true
-enum-map.workspace = true
-serde_json.workspace = true
 
 workspace_hack.workspace = true

libs/pageserver_api/src/models.rs

@@ -1,14 +1,9 @@
-use std::{
-    collections::HashMap,
-    num::{NonZeroU64, NonZeroUsize},
-    time::SystemTime,
-};
+use std::num::{NonZeroU64, NonZeroUsize};
 
 use byteorder::{BigEndian, ReadBytesExt};
 use serde::{Deserialize, Serialize};
 use serde_with::{serde_as, DisplayFromStr};
 use utils::{
-    history_buffer::HistoryBufferWithDropCounter,
     id::{NodeId, TenantId, TimelineId},
     lsn::Lsn,
 };
@@ -34,14 +29,6 @@ pub enum TenantState {
     Broken,
 }
 
-pub mod state {
-    pub const LOADING: &str = "loading";
-    pub const ATTACHING: &str = "attaching";
-    pub const ACTIVE: &str = "active";
-    pub const STOPPING: &str = "stopping";
-    pub const BROKEN: &str = "broken";
-}
-
 impl TenantState {
     pub fn has_in_progress_downloads(&self) -> bool {
         match self {
@@ -52,16 +39,6 @@ impl TenantState {
             Self::Broken => false,
         }
     }
-
-    pub fn as_str(&self) -> &'static str {
-        match self {
-            TenantState::Loading => state::LOADING,
-            TenantState::Attaching => state::ATTACHING,
-            TenantState::Active => state::ACTIVE,
-            TenantState::Stopping => state::STOPPING,
-            TenantState::Broken => state::BROKEN,
-        }
-    }
 }
 
 /// A state of a timeline in pageserver's memory.
@@ -115,11 +92,6 @@ pub struct TenantCreateRequest {
     pub lagging_wal_timeout: Option<String>,
     pub max_lsn_wal_lag: Option<NonZeroU64>,
     pub trace_read_requests: Option<bool>,
-    // We defer the parsing of the eviction_policy field to the request handler.
-    // Otherwise we'd have to move the types for eviction policy into this package.
-    // We might do that once the eviction feature has stabilizied.
-    // For now, this field is not even documented in the openapi_spec.yml.
-    pub eviction_policy: Option<serde_json::Value>,
 }
 
 #[serde_as]
@@ -147,6 +119,7 @@ pub struct TenantConfigRequest {
     #[serde_as(as = "DisplayFromStr")]
     pub tenant_id: TenantId,
     #[serde(default)]
+    #[serde_as(as = "Option<DisplayFromStr>")]
     pub checkpoint_distance: Option<u64>,
     pub checkpoint_timeout: Option<String>,
     pub compaction_target_size: Option<u64>,
@@ -160,11 +133,6 @@ pub struct TenantConfigRequest {
     pub lagging_wal_timeout: Option<String>,
     pub max_lsn_wal_lag: Option<NonZeroU64>,
     pub trace_read_requests: Option<bool>,
-    // We defer the parsing of the eviction_policy field to the request handler.
-    // Otherwise we'd have to move the types for eviction policy into this package.
-    // We might do that once the eviction feature has stabilizied.
-    // For now, this field is not even documented in the openapi_spec.yml.
-    pub eviction_policy: Option<serde_json::Value>,
 }
 
 impl TenantConfigRequest {
@@ -184,7 +152,6 @@ impl TenantConfigRequest {
             lagging_wal_timeout: None,
             max_lsn_wal_lag: None,
             trace_read_requests: None,
-            eviction_policy: None,
         }
     }
 }
@@ -242,130 +209,6 @@ pub struct TimelineInfo {
     pub state: TimelineState,
 }
 
-#[derive(Debug, Clone, Serialize)]
-pub struct LayerMapInfo {
-    pub in_memory_layers: Vec<InMemoryLayerInfo>,
-    pub historic_layers: Vec<HistoricLayerInfo>,
-}
-
-#[derive(Debug, Hash, PartialEq, Eq, Clone, Copy, Serialize, Deserialize, enum_map::Enum)]
-#[repr(usize)]
-pub enum LayerAccessKind {
-    GetValueReconstructData,
-    Iter,
-    KeyIter,
-    Dump,
-}
-
-#[derive(Debug, Clone, Serialize, Deserialize)]
-pub struct LayerAccessStatFullDetails {
-    pub when_millis_since_epoch: u64,
-    pub task_kind: &'static str,
-    pub access_kind: LayerAccessKind,
-}
-
-/// An event that impacts the layer's residence status.
-#[serde_as]
-#[derive(Debug, Clone, Serialize, Deserialize)]
-pub struct LayerResidenceEvent {
-    /// The time when the event occurred.
-    /// NB: this timestamp is captured while the residence status changes.
-    /// So, it might be behind/ahead of the actual residence change by a short amount of time.
-    ///
-    #[serde(rename = "timestamp_millis_since_epoch")]
-    #[serde_as(as = "serde_with::TimestampMilliSeconds")]
-    pub timestamp: SystemTime,
-    /// The new residence status of the layer.
-    pub status: LayerResidenceStatus,
-    /// The reason why we had to record this event.
-    pub reason: LayerResidenceEventReason,
-}
-
-/// The reason for recording a given [`ResidenceEvent`].
-#[derive(Debug, Clone, Copy, Serialize, Deserialize)]
-pub enum LayerResidenceEventReason {
-    /// The layer map is being populated, e.g. during timeline load or attach.
-    /// This includes [`RemoteLayer`] objects created in [`reconcile_with_remote`].
-    /// We need to record such events because there is no persistent storage for the events.
-    LayerLoad,
-    /// We just created the layer (e.g., freeze_and_flush or compaction).
-    /// Such layers are always [`LayerResidenceStatus::Resident`].
-    LayerCreate,
-    /// We on-demand downloaded or evicted the given layer.
-    ResidenceChange,
-}
-
-/// The residence status of the layer, after the given [`LayerResidenceEvent`].
-#[derive(Debug, Clone, Copy, Serialize, Deserialize)]
-pub enum LayerResidenceStatus {
-    /// Residence status for a layer file that exists locally.
-    /// It may also exist on the remote, we don't care here.
-    Resident,
-    /// Residence status for a layer file that only exists on the remote.
-    Evicted,
-}
-
-impl LayerResidenceEvent {
-    pub fn new(status: LayerResidenceStatus, reason: LayerResidenceEventReason) -> Self {
-        Self {
-            status,
-            reason,
-            timestamp: SystemTime::now(),
-        }
-    }
-}
-
-#[derive(Debug, Clone, Serialize)]
-pub struct LayerAccessStats {
-    pub access_count_by_access_kind: HashMap<LayerAccessKind, u64>,
-    pub task_kind_access_flag: Vec<&'static str>,
-    pub first: Option<LayerAccessStatFullDetails>,
-    pub accesses_history: HistoryBufferWithDropCounter<LayerAccessStatFullDetails, 16>,
-    pub residence_events_history: HistoryBufferWithDropCounter<LayerResidenceEvent, 16>,
-}
-
-#[serde_as]
-#[derive(Debug, Clone, Serialize)]
-#[serde(tag = "kind")]
-pub enum InMemoryLayerInfo {
-    Open {
-        #[serde_as(as = "DisplayFromStr")]
-        lsn_start: Lsn,
-    },
-    Frozen {
-        #[serde_as(as = "DisplayFromStr")]
-        lsn_start: Lsn,
-        #[serde_as(as = "DisplayFromStr")]
-        lsn_end: Lsn,
-    },
-}
-
-#[serde_as]
-#[derive(Debug, Clone, Serialize)]
-#[serde(tag = "kind")]
-pub enum HistoricLayerInfo {
-    Delta {
-        layer_file_name: String,
-        layer_file_size: u64,
-
-        #[serde_as(as = "DisplayFromStr")]
-        lsn_start: Lsn,
-        #[serde_as(as = "DisplayFromStr")]
-        lsn_end: Lsn,
-        remote: bool,
-        access_stats: LayerAccessStats,
-    },
-    Image {
-        layer_file_name: String,
-        layer_file_size: u64,
-
-        #[serde_as(as = "DisplayFromStr")]
-        lsn_start: Lsn,
-        remote: bool,
-        access_stats: LayerAccessStats,
-    },
-}
-
 #[derive(Debug, Serialize, Deserialize)]
 pub struct DownloadRemoteLayersTaskSpawnRequest {
     pub max_concurrent_downloads: NonZeroUsize,
@@ -406,7 +249,7 @@ pub struct TimelineGcRequest {
 }
 
 // Wrapped in libpq CopyData
-#[derive(PartialEq, Eq, Debug)]
+#[derive(PartialEq, Eq)]
 pub enum PagestreamFeMessage {
     Exists(PagestreamExistsRequest),
     Nblocks(PagestreamNblocksRequest),

libs/pageserver_api/src/reltag.rs

@@ -98,15 +98,6 @@ impl RelTag {
 
         name
     }
-
-    pub fn with_forknum(&self, forknum: u8) -> Self {
-        RelTag {
-            forknum,
-            spcnode: self.spcnode,
-            dbnode: self.dbnode,
-            relnode: self.relnode,
-        }
-    }
 }
 
 ///

libs/postgres_backend/Cargo.toml

@@ -1,26 +0,0 @@
-[package]
-name = "postgres_backend"
-version = "0.1.0"
-edition.workspace = true
-license.workspace = true
-
-[dependencies]
-async-trait.workspace = true
-anyhow.workspace = true
-bytes.workspace = true
-futures.workspace = true
-rustls.workspace = true
-serde.workspace = true
-thiserror.workspace = true
-tokio.workspace = true
-tokio-rustls.workspace = true
-tracing.workspace = true
-
-pq_proto.workspace = true
-workspace_hack.workspace = true
-
-[dev-dependencies]
-once_cell.workspace = true
-rustls-pemfile.workspace = true
-tokio-postgres.workspace = true
-tokio-postgres-rustls.workspace = true

libs/postgres_backend/src/lib.rs

@@ -1,931 +0,0 @@
-//! Server-side asynchronous Postgres connection, as limited as we need.
-//! To use, create PostgresBackend and run() it, passing the Handler
-//! implementation determining how to process the queries. Currently its API
-//! is rather narrow, but we can extend it once required.
-use anyhow::Context;
-use bytes::Bytes;
-use futures::pin_mut;
-use serde::{Deserialize, Serialize};
-use std::io::ErrorKind;
-use std::net::SocketAddr;
-use std::pin::Pin;
-use std::sync::Arc;
-use std::task::{ready, Poll};
-use std::{fmt, io};
-use std::{future::Future, str::FromStr};
-use tokio::io::{AsyncRead, AsyncWrite};
-use tokio_rustls::TlsAcceptor;
-use tracing::{debug, error, info, trace};
-
-use pq_proto::framed::{ConnectionError, Framed, FramedReader, FramedWriter};
-use pq_proto::{
-    BeMessage, FeMessage, FeStartupPacket, ProtocolError, SQLSTATE_INTERNAL_ERROR,
-    SQLSTATE_SUCCESSFUL_COMPLETION,
-};
-
-/// An error, occurred during query processing:
-/// either during the connection ([`ConnectionError`]) or before/after it.
-#[derive(thiserror::Error, Debug)]
-pub enum QueryError {
-    /// The connection was lost while processing the query.
-    #[error(transparent)]
-    Disconnected(#[from] ConnectionError),
-    /// Some other error
-    #[error(transparent)]
-    Other(#[from] anyhow::Error),
-}
-
-impl From<io::Error> for QueryError {
-    fn from(e: io::Error) -> Self {
-        Self::Disconnected(ConnectionError::Io(e))
-    }
-}
-
-impl QueryError {
-    pub fn pg_error_code(&self) -> &'static [u8; 5] {
-        match self {
-            Self::Disconnected(_) => b"08006",         // connection failure
-            Self::Other(_) => SQLSTATE_INTERNAL_ERROR, // internal error
-        }
-    }
-}
-
-pub fn is_expected_io_error(e: &io::Error) -> bool {
-    use io::ErrorKind::*;
-    matches!(
-        e.kind(),
-        ConnectionRefused | ConnectionAborted | ConnectionReset
-    )
-}
-
-#[async_trait::async_trait]
-pub trait Handler<IO> {
-    /// Handle single query.
-    /// postgres_backend will issue ReadyForQuery after calling this (this
-    /// might be not what we want after CopyData streaming, but currently we don't
-    /// care). It will also flush out the output buffer.
-    async fn process_query(
-        &mut self,
-        pgb: &mut PostgresBackend<IO>,
-        query_string: &str,
-    ) -> Result<(), QueryError>;
-
-    /// Called on startup packet receival, allows to process params.
-    ///
-    /// If Ok(false) is returned postgres_backend will skip auth -- that is needed for new users
-    /// creation is the proxy code. That is quite hacky and ad-hoc solution, may be we could allow
-    /// to override whole init logic in implementations.
-    fn startup(
-        &mut self,
-        _pgb: &mut PostgresBackend<IO>,
-        _sm: &FeStartupPacket,
-    ) -> Result<(), QueryError> {
-        Ok(())
-    }
-
-    /// Check auth jwt
-    fn check_auth_jwt(
-        &mut self,
-        _pgb: &mut PostgresBackend<IO>,
-        _jwt_response: &[u8],
-    ) -> Result<(), QueryError> {
-        Err(QueryError::Other(anyhow::anyhow!("JWT auth failed")))
-    }
-}
-
-/// PostgresBackend protocol state.
-/// XXX: The order of the constructors matters.
-#[derive(Clone, Copy, PartialEq, Eq, PartialOrd)]
-pub enum ProtoState {
-    /// Nothing happened yet.
-    Initialization,
-    /// Encryption handshake is done; waiting for encrypted Startup message.
-    Encrypted,
-    /// Waiting for password (auth token).
-    Authentication,
-    /// Performed handshake and auth, ReadyForQuery is issued.
-    Established,
-    Closed,
-}
-
-#[derive(Clone, Copy)]
-pub enum ProcessMsgResult {
-    Continue,
-    Break,
-}
-
-/// Either plain TCP stream or encrypted one, implementing AsyncRead + AsyncWrite.
-pub enum MaybeTlsStream<IO> {
-    Unencrypted(IO),
-    Tls(Box<tokio_rustls::server::TlsStream<IO>>),
-}
-
-impl<IO: AsyncRead + AsyncWrite + Unpin> AsyncWrite for MaybeTlsStream<IO> {
-    fn poll_write(
-        self: Pin<&mut Self>,
-        cx: &mut std::task::Context<'_>,
-        buf: &[u8],
-    ) -> Poll<io::Result<usize>> {
-        match self.get_mut() {
-            Self::Unencrypted(stream) => Pin::new(stream).poll_write(cx, buf),
-            Self::Tls(stream) => Pin::new(stream).poll_write(cx, buf),
-        }
-    }
-    fn poll_flush(self: Pin<&mut Self>, cx: &mut std::task::Context<'_>) -> Poll<io::Result<()>> {
-        match self.get_mut() {
-            Self::Unencrypted(stream) => Pin::new(stream).poll_flush(cx),
-            Self::Tls(stream) => Pin::new(stream).poll_flush(cx),
-        }
-    }
-    fn poll_shutdown(
-        self: Pin<&mut Self>,
-        cx: &mut std::task::Context<'_>,
-    ) -> Poll<io::Result<()>> {
-        match self.get_mut() {
-            Self::Unencrypted(stream) => Pin::new(stream).poll_shutdown(cx),
-            Self::Tls(stream) => Pin::new(stream).poll_shutdown(cx),
-        }
-    }
-}
-impl<IO: AsyncRead + AsyncWrite + Unpin> AsyncRead for MaybeTlsStream<IO> {
-    fn poll_read(
-        self: Pin<&mut Self>,
-        cx: &mut std::task::Context<'_>,
-        buf: &mut tokio::io::ReadBuf<'_>,
-    ) -> Poll<io::Result<()>> {
-        match self.get_mut() {
-            Self::Unencrypted(stream) => Pin::new(stream).poll_read(cx, buf),
-            Self::Tls(stream) => Pin::new(stream).poll_read(cx, buf),
-        }
-    }
-}
-
-#[derive(Debug, PartialEq, Eq, Clone, Copy, Serialize, Deserialize)]
-pub enum AuthType {
-    Trust,
-    // This mimics postgres's AuthenticationCleartextPassword but instead of password expects JWT
-    NeonJWT,
-}
-
-impl FromStr for AuthType {
-    type Err = anyhow::Error;
-
-    fn from_str(s: &str) -> Result<Self, Self::Err> {
-        match s {
-            "Trust" => Ok(Self::Trust),
-            "NeonJWT" => Ok(Self::NeonJWT),
-            _ => anyhow::bail!("invalid value \"{s}\" for auth type"),
-        }
-    }
-}
-
-impl fmt::Display for AuthType {
-    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
-        f.write_str(match self {
-            AuthType::Trust => "Trust",
-            AuthType::NeonJWT => "NeonJWT",
-        })
-    }
-}
-
-/// Either full duplex Framed or write only half; the latter is left in
-/// PostgresBackend after call to `split`. In principle we could always store a
-/// pair of splitted handles, but that would force to to pay splitting price
-/// (Arc and kinda mutex inside polling) for all uses (e.g. pageserver).
-enum MaybeWriteOnly<IO> {
-    Full(Framed<MaybeTlsStream<IO>>),
-    WriteOnly(FramedWriter<MaybeTlsStream<IO>>),
-    Broken, // temporary value palmed off during the split
-}
-
-impl<IO: AsyncRead + AsyncWrite + Unpin> MaybeWriteOnly<IO> {
-    async fn read_startup_message(&mut self) -> Result<Option<FeStartupPacket>, ConnectionError> {
-        match self {
-            MaybeWriteOnly::Full(framed) => framed.read_startup_message().await,
-            MaybeWriteOnly::WriteOnly(_) => {
-                Err(io::Error::new(ErrorKind::Other, "reading from write only half").into())
-            }
-            MaybeWriteOnly::Broken => panic!("IO on invalid MaybeWriteOnly"),
-        }
-    }
-
-    async fn read_message(&mut self) -> Result<Option<FeMessage>, ConnectionError> {
-        match self {
-            MaybeWriteOnly::Full(framed) => framed.read_message().await,
-            MaybeWriteOnly::WriteOnly(_) => {
-                Err(io::Error::new(ErrorKind::Other, "reading from write only half").into())
-            }
-            MaybeWriteOnly::Broken => panic!("IO on invalid MaybeWriteOnly"),
-        }
-    }
-
-    fn write_message_noflush(&mut self, msg: &BeMessage<'_>) -> Result<(), ProtocolError> {
-        match self {
-            MaybeWriteOnly::Full(framed) => framed.write_message(msg),
-            MaybeWriteOnly::WriteOnly(framed_writer) => framed_writer.write_message_noflush(msg),
-            MaybeWriteOnly::Broken => panic!("IO on invalid MaybeWriteOnly"),
-        }
-    }
-
-    async fn flush(&mut self) -> io::Result<()> {
-        match self {
-            MaybeWriteOnly::Full(framed) => framed.flush().await,
-            MaybeWriteOnly::WriteOnly(framed_writer) => framed_writer.flush().await,
-            MaybeWriteOnly::Broken => panic!("IO on invalid MaybeWriteOnly"),
-        }
-    }
-
-    async fn shutdown(&mut self) -> io::Result<()> {
-        match self {
-            MaybeWriteOnly::Full(framed) => framed.shutdown().await,
-            MaybeWriteOnly::WriteOnly(framed_writer) => framed_writer.shutdown().await,
-            MaybeWriteOnly::Broken => panic!("IO on invalid MaybeWriteOnly"),
-        }
-    }
-}
-
-pub struct PostgresBackend<IO> {
-    framed: MaybeWriteOnly<IO>,
-
-    pub state: ProtoState,
-
-    auth_type: AuthType,
-
-    peer_addr: SocketAddr,
-    pub tls_config: Option<Arc<rustls::ServerConfig>>,
-}
-
-pub type PostgresBackendTCP = PostgresBackend<tokio::net::TcpStream>;
-
-pub fn query_from_cstring(query_string: Bytes) -> Vec<u8> {
-    let mut query_string = query_string.to_vec();
-    if let Some(ch) = query_string.last() {
-        if *ch == 0 {
-            query_string.pop();
-        }
-    }
-    query_string
-}
-
-/// Cast a byte slice to a string slice, dropping null terminator if there's one.
-fn cstr_to_str(bytes: &[u8]) -> anyhow::Result<&str> {
-    let without_null = bytes.strip_suffix(&[0]).unwrap_or(bytes);
-    std::str::from_utf8(without_null).map_err(|e| e.into())
-}
-
-impl PostgresBackend<tokio::net::TcpStream> {
-    pub fn new(
-        socket: tokio::net::TcpStream,
-        auth_type: AuthType,
-        tls_config: Option<Arc<rustls::ServerConfig>>,
-    ) -> io::Result<Self> {
-        let peer_addr = socket.peer_addr()?;
-        let stream = MaybeTlsStream::Unencrypted(socket);
-
-        Ok(Self {
-            framed: MaybeWriteOnly::Full(Framed::new(stream)),
-            state: ProtoState::Initialization,
-            auth_type,
-            tls_config,
-            peer_addr,
-        })
-    }
-}
-
-impl<IO: AsyncRead + AsyncWrite + Unpin> PostgresBackend<IO> {
-    pub fn new_from_io(
-        socket: IO,
-        peer_addr: SocketAddr,
-        auth_type: AuthType,
-        tls_config: Option<Arc<rustls::ServerConfig>>,
-    ) -> io::Result<Self> {
-        let stream = MaybeTlsStream::Unencrypted(socket);
-
-        Ok(Self {
-            framed: MaybeWriteOnly::Full(Framed::new(stream)),
-            state: ProtoState::Initialization,
-            auth_type,
-            tls_config,
-            peer_addr,
-        })
-    }
-
-    pub fn get_peer_addr(&self) -> &SocketAddr {
-        &self.peer_addr
-    }
-
-    /// Read full message or return None if connection is cleanly closed with no
-    /// unprocessed data.
-    pub async fn read_message(&mut self) -> Result<Option<FeMessage>, ConnectionError> {
-        if let ProtoState::Closed = self.state {
-            Ok(None)
-        } else {
-            let m = self.framed.read_message().await?;
-            trace!("read msg {:?}", m);
-            Ok(m)
-        }
-    }
-
-    /// Write message into internal output buffer, doesn't flush it. Technically
-    /// error type can be only ProtocolError here (if, unlikely, serialization
-    /// fails), but callers typically wrap it anyway.
-    pub fn write_message_noflush(
-        &mut self,
-        message: &BeMessage<'_>,
-    ) -> Result<&mut Self, ConnectionError> {
-        self.framed.write_message_noflush(message)?;
-        trace!("wrote msg {:?}", message);
-        Ok(self)
-    }
-
-    /// Flush output buffer into the socket.
-    pub async fn flush(&mut self) -> io::Result<()> {
-        self.framed.flush().await
-    }
-
-    /// Polling version of `flush()`, saves the caller need to pin.
-    pub fn poll_flush(
-        &mut self,
-        cx: &mut std::task::Context<'_>,
-    ) -> Poll<Result<(), std::io::Error>> {
-        let flush_fut = self.flush();
-        pin_mut!(flush_fut);
-        flush_fut.poll(cx)
-    }
-
-    /// Write message into internal output buffer and flush it to the stream.
-    pub async fn write_message(
-        &mut self,
-        message: &BeMessage<'_>,
-    ) -> Result<&mut Self, ConnectionError> {
-        self.write_message_noflush(message)?;
-        self.flush().await?;
-        Ok(self)
-    }
-
-    /// Returns an AsyncWrite implementation that wraps all the data written
-    /// to it in CopyData messages, and writes them to the connection
-    ///
-    /// The caller is responsible for sending CopyOutResponse and CopyDone messages.
-    pub fn copyout_writer(&mut self) -> CopyDataWriter<IO> {
-        CopyDataWriter { pgb: self }
-    }
-
-    /// Wrapper for run_message_loop() that shuts down socket when we are done
-    pub async fn run<F, S>(
-        mut self,
-        handler: &mut impl Handler<IO>,
-        shutdown_watcher: F,
-    ) -> Result<(), QueryError>
-    where
-        F: Fn() -> S,
-        S: Future,
-    {
-        let ret = self.run_message_loop(handler, shutdown_watcher).await;
-        // socket might be already closed, e.g. if previously received error,
-        // so ignore result.
-        self.framed.shutdown().await.ok();
-        ret
-    }
-
-    async fn run_message_loop<F, S>(
-        &mut self,
-        handler: &mut impl Handler<IO>,
-        shutdown_watcher: F,
-    ) -> Result<(), QueryError>
-    where
-        F: Fn() -> S,
-        S: Future,
-    {
-        trace!("postgres backend to {:?} started", self.peer_addr);
-
-        tokio::select!(
-            biased;
-
-            _ = shutdown_watcher() => {
-                // We were requested to shut down.
-                tracing::info!("shutdown request received during handshake");
-                return Ok(())
-            },
-
-            result = self.handshake(handler) => {
-                // Handshake complete.
-                result?;
-                if self.state == ProtoState::Closed {
-                    return Ok(()); // EOF during handshake
-                }
-            }
-        );
-
-        // Authentication completed
-        let mut query_string = Bytes::new();
-        while let Some(msg) = tokio::select!(
-            biased;
-            _ = shutdown_watcher() => {
-                // We were requested to shut down.
-                tracing::info!("shutdown request received in run_message_loop");
-                Ok(None)
-            },
-            msg = self.read_message() => { msg },
-        )? {
-            trace!("got message {:?}", msg);
-
-            let result = self.process_message(handler, msg, &mut query_string).await;
-            self.flush().await?;
-            match result? {
-                ProcessMsgResult::Continue => {
-                    self.flush().await?;
-                    continue;
-                }
-                ProcessMsgResult::Break => break,
-            }
-        }
-
-        trace!("postgres backend to {:?} exited", self.peer_addr);
-        Ok(())
-    }
-
-    /// Try to upgrade MaybeTlsStream into actual TLS one, performing handshake.
-    async fn tls_upgrade(
-        src: MaybeTlsStream<IO>,
-        tls_config: Arc<rustls::ServerConfig>,
-    ) -> anyhow::Result<MaybeTlsStream<IO>> {
-        match src {
-            MaybeTlsStream::Unencrypted(s) => {
-                let acceptor = TlsAcceptor::from(tls_config);
-                let tls_stream = acceptor.accept(s).await?;
-                Ok(MaybeTlsStream::Tls(Box::new(tls_stream)))
-            }
-            MaybeTlsStream::Tls(_) => {
-                anyhow::bail!("TLS already started");
-            }
-        }
-    }
-
-    async fn start_tls(&mut self) -> anyhow::Result<()> {
-        // temporary replace stream with fake to cook TLS one, Indiana Jones style
-        match std::mem::replace(&mut self.framed, MaybeWriteOnly::Broken) {
-            MaybeWriteOnly::Full(framed) => {
-                let tls_config = self
-                    .tls_config
-                    .as_ref()
-                    .context("start_tls called without conf")?
-                    .clone();
-                let tls_framed = framed
-                    .map_stream(|s| PostgresBackend::tls_upgrade(s, tls_config))
-                    .await?;
-                // push back ready TLS stream
-                self.framed = MaybeWriteOnly::Full(tls_framed);
-                Ok(())
-            }
-            MaybeWriteOnly::WriteOnly(_) => {
-                anyhow::bail!("TLS upgrade attempt in split state")
-            }
-            MaybeWriteOnly::Broken => panic!("TLS upgrade on framed in invalid state"),
-        }
-    }
-
-    /// Split off owned read part from which messages can be read in different
-    /// task/thread.
-    pub fn split(&mut self) -> anyhow::Result<PostgresBackendReader<IO>> {
-        // temporary replace stream with fake to cook split one, Indiana Jones style
-        match std::mem::replace(&mut self.framed, MaybeWriteOnly::Broken) {
-            MaybeWriteOnly::Full(framed) => {
-                let (reader, writer) = framed.split();
-                self.framed = MaybeWriteOnly::WriteOnly(writer);
-                Ok(PostgresBackendReader(reader))
-            }
-            MaybeWriteOnly::WriteOnly(_) => {
-                anyhow::bail!("PostgresBackend is already split")
-            }
-            MaybeWriteOnly::Broken => panic!("split on framed in invalid state"),
-        }
-    }
-
-    /// Join read part back.
-    pub fn unsplit(&mut self, reader: PostgresBackendReader<IO>) -> anyhow::Result<()> {
-        // temporary replace stream with fake to cook joined one, Indiana Jones style
-        match std::mem::replace(&mut self.framed, MaybeWriteOnly::Broken) {
-            MaybeWriteOnly::Full(_) => {
-                anyhow::bail!("PostgresBackend is not split")
-            }
-            MaybeWriteOnly::WriteOnly(writer) => {
-                let joined = Framed::unsplit(reader.0, writer);
-                self.framed = MaybeWriteOnly::Full(joined);
-                Ok(())
-            }
-            MaybeWriteOnly::Broken => panic!("unsplit on framed in invalid state"),
-        }
-    }
-
-    /// Perform handshake with the client, transitioning to Established.
-    /// In case of EOF during handshake logs this, sets state to Closed and returns Ok(()).
-    async fn handshake(&mut self, handler: &mut impl Handler<IO>) -> Result<(), QueryError> {
-        while self.state < ProtoState::Authentication {
-            match self.framed.read_startup_message().await? {
-                Some(msg) => {
-                    self.process_startup_message(handler, msg).await?;
-                }
-                None => {
-                    trace!(
-                        "postgres backend to {:?} received EOF during handshake",
-                        self.peer_addr
-                    );
-                    self.state = ProtoState::Closed;
-                    return Ok(());
-                }
-            }
-        }
-
-        // Perform auth, if needed.
-        if self.state == ProtoState::Authentication {
-            match self.framed.read_message().await? {
-                Some(FeMessage::PasswordMessage(m)) => {
-                    assert!(self.auth_type == AuthType::NeonJWT);
-
-                    let (_, jwt_response) = m.split_last().context("protocol violation")?;
-
-                    if let Err(e) = handler.check_auth_jwt(self, jwt_response) {
-                        self.write_message_noflush(&BeMessage::ErrorResponse(
-                            &e.to_string(),
-                            Some(e.pg_error_code()),
-                        ))?;
-                        return Err(e);
-                    }
-
-                    self.write_message_noflush(&BeMessage::AuthenticationOk)?
-                        .write_message_noflush(&BeMessage::CLIENT_ENCODING)?
-                        .write_message(&BeMessage::ReadyForQuery)
-                        .await?;
-                    self.state = ProtoState::Established;
-                }
-                Some(m) => {
-                    return Err(QueryError::Other(anyhow::anyhow!(
-                        "Unexpected message {:?} while waiting for handshake",
-                        m
-                    )));
-                }
-                None => {
-                    trace!(
-                        "postgres backend to {:?} received EOF during auth",
-                        self.peer_addr
-                    );
-                    self.state = ProtoState::Closed;
-                    return Ok(());
-                }
-            }
-        }
-
-        Ok(())
-    }
-
-    /// Process startup packet:
-    /// - transition to Established if auth type is trust
-    /// - transition to Authentication if auth type is NeonJWT.
-    /// - or perform TLS handshake -- then need to call this again to receive
-    ///   actual startup packet.
-    async fn process_startup_message(
-        &mut self,
-        handler: &mut impl Handler<IO>,
-        msg: FeStartupPacket,
-    ) -> Result<(), QueryError> {
-        assert!(self.state < ProtoState::Authentication);
-        let have_tls = self.tls_config.is_some();
-        match msg {
-            FeStartupPacket::SslRequest => {
-                debug!("SSL requested");
-
-                self.write_message(&BeMessage::EncryptionResponse(have_tls))
-                    .await?;
-
-                if have_tls {
-                    self.start_tls().await?;
-                    self.state = ProtoState::Encrypted;
-                }
-            }
-            FeStartupPacket::GssEncRequest => {
-                debug!("GSS requested");
-                self.write_message(&BeMessage::EncryptionResponse(false))
-                    .await?;
-            }
-            FeStartupPacket::StartupMessage { .. } => {
-                if have_tls && !matches!(self.state, ProtoState::Encrypted) {
-                    self.write_message(&BeMessage::ErrorResponse("must connect with TLS", None))
-                        .await?;
-                    return Err(QueryError::Other(anyhow::anyhow!(
-                        "client did not connect with TLS"
-                    )));
-                }
-
-                // NB: startup() may change self.auth_type -- we are using that in proxy code
-                // to bypass auth for new users.
-                handler.startup(self, &msg)?;
-
-                match self.auth_type {
-                    AuthType::Trust => {
-                        self.write_message_noflush(&BeMessage::AuthenticationOk)?
-                            .write_message_noflush(&BeMessage::CLIENT_ENCODING)?
-                            .write_message_noflush(&BeMessage::INTEGER_DATETIMES)?
-                            // The async python driver requires a valid server_version
-                            .write_message_noflush(&BeMessage::server_version("14.1"))?
-                            .write_message(&BeMessage::ReadyForQuery)
-                            .await?;
-                        self.state = ProtoState::Established;
-                    }
-                    AuthType::NeonJWT => {
-                        self.write_message(&BeMessage::AuthenticationCleartextPassword)
-                            .await?;
-                        self.state = ProtoState::Authentication;
-                    }
-                }
-            }
-            FeStartupPacket::CancelRequest { .. } => {
-                return Err(QueryError::Other(anyhow::anyhow!(
-                    "Unexpected CancelRequest message during handshake"
-                )));
-            }
-        }
-        Ok(())
-    }
-
-    async fn process_message(
-        &mut self,
-        handler: &mut impl Handler<IO>,
-        msg: FeMessage,
-        unnamed_query_string: &mut Bytes,
-    ) -> Result<ProcessMsgResult, QueryError> {
-        // Allow only startup and password messages during auth. Otherwise client would be able to bypass auth
-        // TODO: change that to proper top-level match of protocol state with separate message handling for each state
-        assert!(self.state == ProtoState::Established);
-
-        match msg {
-            FeMessage::Query(body) => {
-                // remove null terminator
-                let query_string = cstr_to_str(&body)?;
-
-                trace!("got query {query_string:?}");
-                if let Err(e) = handler.process_query(self, query_string).await {
-                    log_query_error(query_string, &e);
-                    let short_error = short_error(&e);
-                    self.write_message_noflush(&BeMessage::ErrorResponse(
-                        &short_error,
-                        Some(e.pg_error_code()),
-                    ))?;
-                }
-                self.write_message_noflush(&BeMessage::ReadyForQuery)?;
-            }
-
-            FeMessage::Parse(m) => {
-                *unnamed_query_string = m.query_string;
-                self.write_message_noflush(&BeMessage::ParseComplete)?;
-            }
-
-            FeMessage::Describe(_) => {
-                self.write_message_noflush(&BeMessage::ParameterDescription)?
-                    .write_message_noflush(&BeMessage::NoData)?;
-            }
-
-            FeMessage::Bind(_) => {
-                self.write_message_noflush(&BeMessage::BindComplete)?;
-            }
-
-            FeMessage::Close(_) => {
-                self.write_message_noflush(&BeMessage::CloseComplete)?;
-            }
-
-            FeMessage::Execute(_) => {
-                let query_string = cstr_to_str(unnamed_query_string)?;
-                trace!("got execute {query_string:?}");
-                if let Err(e) = handler.process_query(self, query_string).await {
-                    log_query_error(query_string, &e);
-                    self.write_message_noflush(&BeMessage::ErrorResponse(
-                        &e.to_string(),
-                        Some(e.pg_error_code()),
-                    ))?;
-                }
-                // NOTE there is no ReadyForQuery message. This handler is used
-                // for basebackup and it uses CopyOut which doesn't require
-                // ReadyForQuery message and backend just switches back to
-                // processing mode after sending CopyDone or ErrorResponse.
-            }
-
-            FeMessage::Sync => {
-                self.write_message_noflush(&BeMessage::ReadyForQuery)?;
-            }
-
-            FeMessage::Terminate => {
-                return Ok(ProcessMsgResult::Break);
-            }
-
-            // We prefer explicit pattern matching to wildcards, because
-            // this helps us spot the places where new variants are missing
-            FeMessage::CopyData(_)
-            | FeMessage::CopyDone
-            | FeMessage::CopyFail
-            | FeMessage::PasswordMessage(_) => {
-                return Err(QueryError::Other(anyhow::anyhow!(
-                    "unexpected message type: {msg:?}",
-                )));
-            }
-        }
-
-        Ok(ProcessMsgResult::Continue)
-    }
-
-    /// Log as info/error result of handling COPY stream and send back
-    /// ErrorResponse if that makes sense. Shutdown the stream if we got
-    /// Terminate. TODO: transition into waiting for Sync msg if we initiate the
-    /// close.
-    pub async fn handle_copy_stream_end(&mut self, end: CopyStreamHandlerEnd) {
-        use CopyStreamHandlerEnd::*;
-
-        let expected_end = match &end {
-            ServerInitiated(_) | CopyDone | CopyFail | Terminate | EOF => true,
-            CopyStreamHandlerEnd::Disconnected(ConnectionError::Io(io_error))
-                if is_expected_io_error(io_error) =>
-            {
-                true
-            }
-            _ => false,
-        };
-        if expected_end {
-            info!("terminated: {:#}", end);
-        } else {
-            error!("terminated: {:?}", end);
-        }
-
-        // Note: no current usages ever send this
-        if let CopyDone = &end {
-            if let Err(e) = self.write_message(&BeMessage::CopyDone).await {
-                error!("failed to send CopyDone: {}", e);
-            }
-        }
-
-        if let Terminate = &end {
-            self.state = ProtoState::Closed;
-        }
-
-        let err_to_send_and_errcode = match &end {
-            ServerInitiated(_) => Some((end.to_string(), SQLSTATE_SUCCESSFUL_COMPLETION)),
-            Other(_) => Some((format!("{end:#}"), SQLSTATE_INTERNAL_ERROR)),
-            // Note: CopyFail in duplex copy is somewhat unexpected (at least to
-            // PG walsender; evidently and per my docs reading client should
-            // finish it with CopyDone). It is not a problem to recover from it
-            // finishing the stream in both directions like we do, but note that
-            // sync rust-postgres client (which we don't use anymore) hangs if
-            // socket is not closed here.
-            // https://github.com/sfackler/rust-postgres/issues/755
-            // https://github.com/neondatabase/neon/issues/935
-            //
-            // Currently, the version of tokio_postgres replication patch we use
-            // sends this when it closes the stream (e.g. pageserver decided to
-            // switch conn to another safekeeper and client gets dropped).
-            // Moreover, seems like 'connection' task errors with 'unexpected
-            // message from server' when it receives ErrorResponse (anything but
-            // CopyData/CopyDone) back.
-            CopyFail => Some((end.to_string(), SQLSTATE_SUCCESSFUL_COMPLETION)),
-            _ => None,
-        };
-        if let Some((err, errcode)) = err_to_send_and_errcode {
-            if let Err(ee) = self
-                .write_message(&BeMessage::ErrorResponse(&err, Some(errcode)))
-                .await
-            {
-                error!("failed to send ErrorResponse: {}", ee);
-            }
-        }
-    }
-}
-
-pub struct PostgresBackendReader<IO>(FramedReader<MaybeTlsStream<IO>>);
-
-impl<IO: AsyncRead + AsyncWrite + Unpin> PostgresBackendReader<IO> {
-    /// Read full message or return None if connection is cleanly closed with no
-    /// unprocessed data.
-    pub async fn read_message(&mut self) -> Result<Option<FeMessage>, ConnectionError> {
-        let m = self.0.read_message().await?;
-        trace!("read msg {:?}", m);
-        Ok(m)
-    }
-
-    /// Get CopyData contents of the next message in COPY stream or error
-    /// closing it. The error type is wider than actual errors which can happen
-    /// here -- it includes 'Other' and 'ServerInitiated', but that's ok for
-    /// current callers.
-    pub async fn read_copy_message(&mut self) -> Result<Bytes, CopyStreamHandlerEnd> {
-        match self.read_message().await? {
-            Some(msg) => match msg {
-                FeMessage::CopyData(m) => Ok(m),
-                FeMessage::CopyDone => Err(CopyStreamHandlerEnd::CopyDone),
-                FeMessage::CopyFail => Err(CopyStreamHandlerEnd::CopyFail),
-                FeMessage::Terminate => Err(CopyStreamHandlerEnd::Terminate),
-                _ => Err(CopyStreamHandlerEnd::from(ConnectionError::Protocol(
-                    ProtocolError::Protocol(format!("unexpected message in COPY stream {:?}", msg)),
-                ))),
-            },
-            None => Err(CopyStreamHandlerEnd::EOF),
-        }
-    }
-}
-
-///
-/// A futures::AsyncWrite implementation that wraps all data written to it in CopyData
-/// messages.
-///
-
-pub struct CopyDataWriter<'a, IO> {
-    pgb: &'a mut PostgresBackend<IO>,
-}
-
-impl<'a, IO: AsyncRead + AsyncWrite + Unpin> AsyncWrite for CopyDataWriter<'a, IO> {
-    fn poll_write(
-        self: Pin<&mut Self>,
-        cx: &mut std::task::Context<'_>,
-        buf: &[u8],
-    ) -> Poll<Result<usize, std::io::Error>> {
-        let this = self.get_mut();
-
-        // It's not strictly required to flush between each message, but makes it easier
-        // to view in wireshark, and usually the messages that the callers write are
-        // decently-sized anyway.
-        if let Err(err) = ready!(this.pgb.poll_flush(cx)) {
-            return Poll::Ready(Err(err));
-        }
-
-        // CopyData
-        // XXX: if the input is large, we should split it into multiple messages.
-        // Not sure what the threshold should be, but the ultimate hard limit is that
-        // the length cannot exceed u32.
-        this.pgb
-            .write_message_noflush(&BeMessage::CopyData(buf))
-            // write_message only writes to the buffer, so it can fail iff the
-            // message is invaid, but CopyData can't be invalid.
-            .map_err(|_| io::Error::new(ErrorKind::Other, "failed to serialize CopyData"))?;
-
-        Poll::Ready(Ok(buf.len()))
-    }
-
-    fn poll_flush(
-        self: Pin<&mut Self>,
-        cx: &mut std::task::Context<'_>,
-    ) -> Poll<Result<(), std::io::Error>> {
-        let this = self.get_mut();
-        this.pgb.poll_flush(cx)
-    }
-
-    fn poll_shutdown(
-        self: Pin<&mut Self>,
-        cx: &mut std::task::Context<'_>,
-    ) -> Poll<Result<(), std::io::Error>> {
-        let this = self.get_mut();
-        this.pgb.poll_flush(cx)
-    }
-}
-
-pub fn short_error(e: &QueryError) -> String {
-    match e {
-        QueryError::Disconnected(connection_error) => connection_error.to_string(),
-        QueryError::Other(e) => format!("{e:#}"),
-    }
-}
-
-fn log_query_error(query: &str, e: &QueryError) {
-    match e {
-        QueryError::Disconnected(ConnectionError::Io(io_error)) => {
-            if is_expected_io_error(io_error) {
-                info!("query handler for '{query}' failed with expected io error: {io_error}");
-            } else {
-                error!("query handler for '{query}' failed with io error: {io_error}");
-            }
-        }
-        QueryError::Disconnected(other_connection_error) => {
-            error!("query handler for '{query}' failed with connection error: {other_connection_error:?}")
-        }
-        QueryError::Other(e) => {
-            error!("query handler for '{query}' failed: {e:?}");
-        }
-    }
-}
-
-/// Something finishing handling of COPY stream, see handle_copy_stream_end.
-/// This is not always a real error, but it allows to use ? and thiserror impls.
-#[derive(thiserror::Error, Debug)]
-pub enum CopyStreamHandlerEnd {
-    /// Handler initiates the end of streaming.
-    #[error("{0}")]
-    ServerInitiated(String),
-    #[error("received CopyDone")]
-    CopyDone,
-    #[error("received CopyFail")]
-    CopyFail,
-    #[error("received Terminate")]
-    Terminate,
-    #[error("EOF on COPY stream")]
-    EOF,
-    /// The connection was lost
-    #[error(transparent)]
-    Disconnected(#[from] ConnectionError),
-    /// Some other error
-    #[error(transparent)]
-    Other(#[from] anyhow::Error),
-}

libs/postgres_backend/tests/simple_select.rs

@@ -1,140 +0,0 @@
-/// Test postgres_backend_async with tokio_postgres
-use once_cell::sync::Lazy;
-use postgres_backend::{AuthType, Handler, PostgresBackend, QueryError};
-use pq_proto::{BeMessage, RowDescriptor};
-use std::io::Cursor;
-use std::{future, sync::Arc};
-use tokio::io::{AsyncRead, AsyncWrite};
-use tokio::net::{TcpListener, TcpStream};
-use tokio_postgres::config::SslMode;
-use tokio_postgres::tls::MakeTlsConnect;
-use tokio_postgres::{Config, NoTls, SimpleQueryMessage};
-use tokio_postgres_rustls::MakeRustlsConnect;
-
-// generate client, server test streams
-async fn make_tcp_pair() -> (TcpStream, TcpStream) {
-    let listener = TcpListener::bind("127.0.0.1:0").await.unwrap();
-    let addr = listener.local_addr().unwrap();
-    let client_stream = TcpStream::connect(addr).await.unwrap();
-    let (server_stream, _) = listener.accept().await.unwrap();
-    (client_stream, server_stream)
-}
-
-struct TestHandler {}
-
-#[async_trait::async_trait]
-impl<IO: AsyncRead + AsyncWrite + Unpin + Send> Handler<IO> for TestHandler {
-    // return single col 'hey' for any query
-    async fn process_query(
-        &mut self,
-        pgb: &mut PostgresBackend<IO>,
-        _query_string: &str,
-    ) -> Result<(), QueryError> {
-        pgb.write_message_noflush(&BeMessage::RowDescription(&[RowDescriptor::text_col(
-            b"hey",
-        )]))?
-        .write_message_noflush(&BeMessage::DataRow(&[Some("hey".as_bytes())]))?
-        .write_message_noflush(&BeMessage::CommandComplete(b"SELECT 1"))?;
-        Ok(())
-    }
-}
-
-// test that basic select works
-#[tokio::test]
-async fn simple_select() {
-    let (client_sock, server_sock) = make_tcp_pair().await;
-
-    // create and run pgbackend
-    let pgbackend =
-        PostgresBackend::new(server_sock, AuthType::Trust, None).expect("pgbackend creation");
-
-    tokio::spawn(async move {
-        let mut handler = TestHandler {};
-        pgbackend.run(&mut handler, future::pending::<()>).await
-    });
-
-    let conf = Config::new();
-    let (client, connection) = conf.connect_raw(client_sock, NoTls).await.expect("connect");
-    // The connection object performs the actual communication with the database,
-    // so spawn it off to run on its own.
-    tokio::spawn(async move {
-        if let Err(e) = connection.await {
-            eprintln!("connection error: {}", e);
-        }
-    });
-
-    let first_val = &(client.simple_query("SELECT 42;").await.expect("select"))[0];
-    if let SimpleQueryMessage::Row(row) = first_val {
-        let first_col = row.get(0).expect("first column");
-        assert_eq!(first_col, "hey");
-    } else {
-        panic!("expected SimpleQueryMessage::Row");
-    }
-}
-
-static KEY: Lazy<rustls::PrivateKey> = Lazy::new(|| {
-    let mut cursor = Cursor::new(include_bytes!("key.pem"));
-    rustls::PrivateKey(rustls_pemfile::rsa_private_keys(&mut cursor).unwrap()[0].clone())
-});
-
-static CERT: Lazy<rustls::Certificate> = Lazy::new(|| {
-    let mut cursor = Cursor::new(include_bytes!("cert.pem"));
-    rustls::Certificate(rustls_pemfile::certs(&mut cursor).unwrap()[0].clone())
-});
-
-// test that basic select with ssl works
-#[tokio::test]
-async fn simple_select_ssl() {
-    let (client_sock, server_sock) = make_tcp_pair().await;
-
-    let server_cfg = rustls::ServerConfig::builder()
-        .with_safe_defaults()
-        .with_no_client_auth()
-        .with_single_cert(vec![CERT.clone()], KEY.clone())
-        .unwrap();
-    let tls_config = Some(Arc::new(server_cfg));
-    let pgbackend =
-        PostgresBackend::new(server_sock, AuthType::Trust, tls_config).expect("pgbackend creation");
-
-    tokio::spawn(async move {
-        let mut handler = TestHandler {};
-        pgbackend.run(&mut handler, future::pending::<()>).await
-    });
-
-    let client_cfg = rustls::ClientConfig::builder()
-        .with_safe_defaults()
-        .with_root_certificates({
-            let mut store = rustls::RootCertStore::empty();
-            store.add(&CERT).unwrap();
-            store
-        })
-        .with_no_client_auth();
-    let mut make_tls_connect = tokio_postgres_rustls::MakeRustlsConnect::new(client_cfg);
-    let tls_connect = <MakeRustlsConnect as MakeTlsConnect<TcpStream>>::make_tls_connect(
-        &mut make_tls_connect,
-        "localhost",
-    )
-    .expect("make_tls_connect");
-
-    let mut conf = Config::new();
-    conf.ssl_mode(SslMode::Require);
-    let (client, connection) = conf
-        .connect_raw(client_sock, tls_connect)
-        .await
-        .expect("connect");
-    // The connection object performs the actual communication with the database,
-    // so spawn it off to run on its own.
-    tokio::spawn(async move {
-        if let Err(e) = connection.await {
-            eprintln!("connection error: {}", e);
-        }
-    });
-
-    let first_val = &(client.simple_query("SELECT 42;").await.expect("select"))[0];
-    if let SimpleQueryMessage::Row(row) = first_val {
-        let first_col = row.get(0).expect("first column");
-        assert_eq!(first_col, "hey");
-    } else {
-        panic!("expected SimpleQueryMessage::Row");
-    }
-}

libs/postgres_ffi/build.rs

@@ -63,7 +63,10 @@ fn main() -> anyhow::Result<()> {
             pg_install_dir_versioned = cwd.join("..").join("..").join(pg_install_dir_versioned);
         }
 
-        let pg_config_bin = pg_install_dir_versioned.join("bin").join("pg_config");
+        let pg_config_bin = pg_install_dir_versioned
+            .join(pg_version)
+            .join("bin")
+            .join("pg_config");
         let inc_server_path: String = if pg_config_bin.exists() {
             let output = Command::new(pg_config_bin)
                 .arg("--includedir-server")

libs/pq_proto/Cargo.toml

@@ -5,8 +5,8 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
+anyhow.workspace = true
 bytes.workspace = true
-byteorder.workspace = true
 pin-project-lite.workspace = true
 postgres-protocol.workspace = true
 rand.workspace = true

libs/pq_proto/src/framed.rs

@@ -1,244 +0,0 @@
-//! Provides `Framed` -- writing/flushing and reading Postgres messages to/from
-//! the async stream based on (and buffered with) BytesMut. All functions are
-//! cancellation safe.
-//!
-//! It is similar to what tokio_util::codec::Framed with appropriate codec
-//! provides, but `FramedReader` and `FramedWriter` read/write parts can be used
-//! separately without using split from futures::stream::StreamExt (which
-//! allocates box[1] in polling internally). tokio::io::split is used for splitting
-//! instead. Plus we customize error messages more than a single type for all io
-//! calls.
-//!
-//! [1] https://docs.rs/futures-util/0.3.26/src/futures_util/lock/bilock.rs.html#107
-use bytes::{Buf, BytesMut};
-use std::{
-    future::Future,
-    io::{self, ErrorKind},
-};
-use tokio::io::{AsyncRead, AsyncReadExt, AsyncWrite, AsyncWriteExt, ReadHalf, WriteHalf};
-
-use crate::{BeMessage, FeMessage, FeStartupPacket, ProtocolError};
-
-const INITIAL_CAPACITY: usize = 8 * 1024;
-
-/// Error on postgres connection: either IO (physical transport error) or
-/// protocol violation.
-#[derive(thiserror::Error, Debug)]
-pub enum ConnectionError {
-    #[error(transparent)]
-    Io(#[from] io::Error),
-    #[error(transparent)]
-    Protocol(#[from] ProtocolError),
-}
-
-impl ConnectionError {
-    /// Proxy stream.rs uses only io::Error; provide it.
-    pub fn into_io_error(self) -> io::Error {
-        match self {
-            ConnectionError::Io(io) => io,
-            ConnectionError::Protocol(pe) => io::Error::new(io::ErrorKind::Other, pe.to_string()),
-        }
-    }
-}
-
-/// Wraps async io `stream`, providing messages to write/flush + read Postgres
-/// messages.
-pub struct Framed<S> {
-    stream: S,
-    read_buf: BytesMut,
-    write_buf: BytesMut,
-}
-
-impl<S> Framed<S> {
-    pub fn new(stream: S) -> Self {
-        Self {
-            stream,
-            read_buf: BytesMut::with_capacity(INITIAL_CAPACITY),
-            write_buf: BytesMut::with_capacity(INITIAL_CAPACITY),
-        }
-    }
-
-    /// Get a shared reference to the underlying stream.
-    pub fn get_ref(&self) -> &S {
-        &self.stream
-    }
-
-    /// Deconstruct into the underlying stream and read buffer.
-    pub fn into_inner(self) -> (S, BytesMut) {
-        (self.stream, self.read_buf)
-    }
-
-    /// Return new Framed with stream type transformed by async f, for TLS
-    /// upgrade.
-    pub async fn map_stream<S2, E, F, Fut>(self, f: F) -> Result<Framed<S2>, E>
-    where
-        F: FnOnce(S) -> Fut,
-        Fut: Future<Output = Result<S2, E>>,
-    {
-        let stream = f(self.stream).await?;
-        Ok(Framed {
-            stream,
-            read_buf: self.read_buf,
-            write_buf: self.write_buf,
-        })
-    }
-}
-
-impl<S: AsyncRead + Unpin> Framed<S> {
-    pub async fn read_startup_message(
-        &mut self,
-    ) -> Result<Option<FeStartupPacket>, ConnectionError> {
-        read_message(&mut self.stream, &mut self.read_buf, FeStartupPacket::parse).await
-    }
-
-    pub async fn read_message(&mut self) -> Result<Option<FeMessage>, ConnectionError> {
-        read_message(&mut self.stream, &mut self.read_buf, FeMessage::parse).await
-    }
-}
-
-impl<S: AsyncWrite + Unpin> Framed<S> {
-    /// Write next message to the output buffer; doesn't flush.
-    pub fn write_message(&mut self, msg: &BeMessage<'_>) -> Result<(), ProtocolError> {
-        BeMessage::write(&mut self.write_buf, msg)
-    }
-
-    /// Flush out the buffer. This function is cancellation safe: it can be
-    /// interrupted and flushing will be continued in the next call.
-    pub async fn flush(&mut self) -> Result<(), io::Error> {
-        flush(&mut self.stream, &mut self.write_buf).await
-    }
-
-    /// Flush out the buffer and shutdown the stream.
-    pub async fn shutdown(&mut self) -> Result<(), io::Error> {
-        shutdown(&mut self.stream, &mut self.write_buf).await
-    }
-}
-
-impl<S: AsyncRead + AsyncWrite + Unpin> Framed<S> {
-    /// Split into owned read and write parts. Beware of potential issues with
-    /// using halves in different tasks on TLS stream:
-    /// https://github.com/tokio-rs/tls/issues/40
-    pub fn split(self) -> (FramedReader<S>, FramedWriter<S>) {
-        let (read_half, write_half) = tokio::io::split(self.stream);
-        let reader = FramedReader {
-            stream: read_half,
-            read_buf: self.read_buf,
-        };
-        let writer = FramedWriter {
-            stream: write_half,
-            write_buf: self.write_buf,
-        };
-        (reader, writer)
-    }
-
-    /// Join read and write parts back.
-    pub fn unsplit(reader: FramedReader<S>, writer: FramedWriter<S>) -> Self {
-        Self {
-            stream: reader.stream.unsplit(writer.stream),
-            read_buf: reader.read_buf,
-            write_buf: writer.write_buf,
-        }
-    }
-}
-
-/// Read-only version of `Framed`.
-pub struct FramedReader<S> {
-    stream: ReadHalf<S>,
-    read_buf: BytesMut,
-}
-
-impl<S: AsyncRead + Unpin> FramedReader<S> {
-    pub async fn read_message(&mut self) -> Result<Option<FeMessage>, ConnectionError> {
-        read_message(&mut self.stream, &mut self.read_buf, FeMessage::parse).await
-    }
-}
-
-/// Write-only version of `Framed`.
-pub struct FramedWriter<S> {
-    stream: WriteHalf<S>,
-    write_buf: BytesMut,
-}
-
-impl<S: AsyncWrite + Unpin> FramedWriter<S> {
-    /// Write next message to the output buffer; doesn't flush.
-    pub fn write_message_noflush(&mut self, msg: &BeMessage<'_>) -> Result<(), ProtocolError> {
-        BeMessage::write(&mut self.write_buf, msg)
-    }
-
-    /// Flush out the buffer. This function is cancellation safe: it can be
-    /// interrupted and flushing will be continued in the next call.
-    pub async fn flush(&mut self) -> Result<(), io::Error> {
-        flush(&mut self.stream, &mut self.write_buf).await
-    }
-
-    /// Flush out the buffer and shutdown the stream.
-    pub async fn shutdown(&mut self) -> Result<(), io::Error> {
-        shutdown(&mut self.stream, &mut self.write_buf).await
-    }
-}
-
-/// Read next message from the stream. Returns Ok(None), if EOF happened and we
-/// don't have remaining data in the buffer. This function is cancellation safe:
-/// you can drop future which is not yet complete and finalize reading message
-/// with the next call.
-///
-/// Parametrized to allow reading startup or usual message, having different
-/// format.
-async fn read_message<S: AsyncRead + Unpin, M, P>(
-    stream: &mut S,
-    read_buf: &mut BytesMut,
-    parse: P,
-) -> Result<Option<M>, ConnectionError>
-where
-    P: Fn(&mut BytesMut) -> Result<Option<M>, ProtocolError>,
-{
-    loop {
-        if let Some(msg) = parse(read_buf)? {
-            return Ok(Some(msg));
-        }
-        // If we can't build a frame yet, try to read more data and try again.
-        // Make sure we've got room for at least one byte to read to ensure
-        // that we don't get a spurious 0 that looks like EOF.
-        read_buf.reserve(1);
-        if stream.read_buf(read_buf).await? == 0 {
-            if read_buf.has_remaining() {
-                return Err(io::Error::new(
-                    ErrorKind::UnexpectedEof,
-                    "EOF with unprocessed data in the buffer",
-                )
-                .into());
-            } else {
-                return Ok(None); // clean EOF
-            }
-        }
-    }
-}
-
-async fn flush<S: AsyncWrite + Unpin>(
-    stream: &mut S,
-    write_buf: &mut BytesMut,
-) -> Result<(), io::Error> {
-    while write_buf.has_remaining() {
-        let bytes_written = stream.write(write_buf.chunk()).await?;
-        if bytes_written == 0 {
-            return Err(io::Error::new(
-                ErrorKind::WriteZero,
-                "failed to write message",
-            ));
-        }
-        // The advanced part will be garbage collected, likely during shifting
-        // data left on next attempt to write to buffer when free space is not
-        // enough.
-        write_buf.advance(bytes_written);
-    }
-    write_buf.clear();
-    stream.flush().await
-}
-
-async fn shutdown<S: AsyncWrite + Unpin>(
-    stream: &mut S,
-    write_buf: &mut BytesMut,
-) -> Result<(), io::Error> {
-    flush(stream, write_buf).await?;
-    stream.shutdown().await
-}

libs/pq_proto/src/lib.rs

@@ -2,18 +2,24 @@
 //! <https://www.postgresql.org/docs/devel/protocol-message-formats.html>
 //! on message formats.
 
-pub mod framed;
+// Tools for calling certain async methods in sync contexts.
+pub mod sync;
 
-use byteorder::{BigEndian, ReadBytesExt};
+use anyhow::{ensure, Context, Result};
 use bytes::{Buf, BufMut, Bytes, BytesMut};
 use postgres_protocol::PG_EPOCH;
 use serde::{Deserialize, Serialize};
 use std::{
     borrow::Cow,
     collections::HashMap,
-    fmt, io, str,
+    fmt,
+    future::Future,
+    io::{self, Cursor},
+    str,
     time::{Duration, SystemTime},
 };
+use sync::{AsyncishRead, SyncFuture};
+use tokio::io::AsyncReadExt;
 use tracing::{trace, warn};
 
 pub type Oid = u32;
@@ -25,6 +31,7 @@ pub const TEXT_OID: Oid = 25;
 
 #[derive(Debug)]
 pub enum FeMessage {
+    StartupPacket(FeStartupPacket),
     // Simple query.
     Query(Bytes),
     // Extended query protocol.
@@ -68,36 +75,27 @@ impl StartupMessageParams {
     /// taking into account all escape sequences but leaving them as-is.
     /// [`None`] means that there's no `options` in [`Self`].
     pub fn options_raw(&self) -> Option<impl Iterator<Item = &str>> {
-        self.get("options").map(Self::parse_options_raw)
-    }
-
-    /// Split command-line options according to PostgreSQL's logic,
-    /// applying all escape sequences (using owned strings as needed).
-    /// [`None`] means that there's no `options` in [`Self`].
-    pub fn options_escaped(&self) -> Option<impl Iterator<Item = Cow<'_, str>>> {
-        self.get("options").map(Self::parse_options_escaped)
-    }
-
-    /// Split command-line options according to PostgreSQL's logic,
-    /// taking into account all escape sequences but leaving them as-is.
-    pub fn parse_options_raw(input: &str) -> impl Iterator<Item = &str> {
         // See `postgres: pg_split_opts`.
         let mut last_was_escape = false;
-        input
+        let iter = self
+            .get("options")?
             .split(move |c: char| {
                 // We split by non-escaped whitespace symbols.
                 let should_split = c.is_ascii_whitespace() && !last_was_escape;
                 last_was_escape = c == '\\' && !last_was_escape;
                 should_split
             })
-            .filter(|s| !s.is_empty())
+            .filter(|s| !s.is_empty());
+
+        Some(iter)
     }
 
     /// Split command-line options according to PostgreSQL's logic,
     /// applying all escape sequences (using owned strings as needed).
-    pub fn parse_options_escaped(input: &str) -> impl Iterator<Item = Cow<'_, str>> {
+    /// [`None`] means that there's no `options` in [`Self`].
+    pub fn options_escaped(&self) -> Option<impl Iterator<Item = Cow<'_, str>>> {
         // See `postgres: pg_split_opts`.
-        Self::parse_options_raw(input).map(|s| {
+        let iter = self.options_raw()?.map(|s| {
             let mut preserve_next_escape = false;
             let escape = |c| {
                 // We should remove '\\' unless it's preceded by '\\'.
@@ -110,12 +108,9 @@ impl StartupMessageParams {
                 true => Cow::Owned(s.replace(escape, "")),
                 false => Cow::Borrowed(s),
             }
-        })
-    }
+        });
 
-    /// Iterate through key-value pairs in an arbitrary order.
-    pub fn iter(&self) -> impl Iterator<Item = (&str, &str)> {
-        self.params.iter().map(|(k, v)| (k.as_str(), v.as_str()))
+        Some(iter)
     }
 
     // This function is mostly useful in tests.
@@ -184,205 +179,260 @@ pub struct FeExecuteMessage {
 #[derive(Debug)]
 pub struct FeCloseMessage;
 
-/// An error occured while parsing or serializing raw stream into Postgres
-/// messages.
-#[derive(thiserror::Error, Debug)]
-pub enum ProtocolError {
-    /// Invalid packet was received from the client (e.g. unexpected message
-    /// type or broken len).
-    #[error("Protocol error: {0}")]
-    Protocol(String),
-    /// Failed to parse or, (unlikely), serialize a protocol message.
-    #[error("Message parse error: {0}")]
-    BadMessage(String),
+/// Retry a read on EINTR
+///
+/// This runs the enclosed expression, and if it returns
+/// Err(io::ErrorKind::Interrupted), retries it.
+macro_rules! retry_read {
+    ( $x:expr ) => {
+        loop {
+            match $x {
+                Err(e) if e.kind() == io::ErrorKind::Interrupted => continue,
+                res => break res,
+            }
+        }
+    };
 }
 
-impl ProtocolError {
-    /// Proxy stream.rs uses only io::Error; provide it.
+/// An error occured during connection being open.
+#[derive(thiserror::Error, Debug)]
+pub enum ConnectionError {
+    /// IO error during writing to or reading from the connection socket.
+    #[error("Socket IO error: {0}")]
+    Socket(std::io::Error),
+    /// Invalid packet was received from client
+    #[error("Protocol error: {0}")]
+    Protocol(String),
+    /// Failed to parse a protocol mesage
+    #[error("Message parse error: {0}")]
+    MessageParse(anyhow::Error),
+}
+
+impl From<anyhow::Error> for ConnectionError {
+    fn from(e: anyhow::Error) -> Self {
+        Self::MessageParse(e)
+    }
+}
+
+impl ConnectionError {
     pub fn into_io_error(self) -> io::Error {
-        io::Error::new(io::ErrorKind::Other, self.to_string())
+        match self {
+            ConnectionError::Socket(io) => io,
+            other => io::Error::new(io::ErrorKind::Other, other.to_string()),
+        }
     }
 }
 
 impl FeMessage {
-    /// Read and parse one message from the `buf` input buffer. If there is at
-    /// least one valid message, returns it, advancing `buf`; redundant copies
-    /// are avoided, as thanks to `bytes` crate ptrs in parsed message point
-    /// directly into the `buf` (processed data is garbage collected after
-    /// parsed message is dropped).
+    /// Read one message from the stream.
+    /// This function returns `Ok(None)` in case of EOF.
+    /// One way to handle this properly:
     ///
-    /// Returns None if `buf` doesn't contain enough data for a single message.
-    /// For efficiency, tries to reserve large enough space in `buf` for the
-    /// next message in this case to save the repeated calls.
+    /// ```
+    /// # use std::io;
+    /// # use pq_proto::FeMessage;
+    /// #
+    /// # fn process_message(msg: FeMessage) -> anyhow::Result<()> {
+    /// #     Ok(())
+    /// # };
+    /// #
+    /// fn do_the_job(stream: &mut (impl io::Read + Unpin)) -> anyhow::Result<()> {
+    ///     while let Some(msg) = FeMessage::read(stream)? {
+    ///         process_message(msg)?;
+    ///     }
     ///
-    /// Returns Error if message is malformed, the only possible ErrorKind is
-    /// InvalidInput.
-    //
-    // Inspired by rust-postgres Message::parse.
-    pub fn parse(buf: &mut BytesMut) -> Result<Option<FeMessage>, ProtocolError> {
-        // Every message contains message type byte and 4 bytes len; can't do
-        // much without them.
-        if buf.len() < 5 {
-            let to_read = 5 - buf.len();
-            buf.reserve(to_read);
-            return Ok(None);
-        }
+    ///     Ok(())
+    /// }
+    /// ```
+    #[inline(never)]
+    pub fn read(
+        stream: &mut (impl io::Read + Unpin),
+    ) -> Result<Option<FeMessage>, ConnectionError> {
+        Self::read_fut(&mut AsyncishRead(stream)).wait()
+    }
 
-        // We shouldn't advance `buf` as probably full message is not there yet,
-        // so can't directly use Bytes::get_u32 etc.
-        let tag = buf[0];
-        let len = (&buf[1..5]).read_u32::<BigEndian>().unwrap();
-        if len < 4 {
-            return Err(ProtocolError::Protocol(format!(
-                "invalid message length {}",
-                len
-            )));
-        }
+    /// Read one message from the stream.
+    /// See documentation for `Self::read`.
+    pub fn read_fut<Reader>(
+        stream: &mut Reader,
+    ) -> SyncFuture<Reader, impl Future<Output = Result<Option<FeMessage>, ConnectionError>> + '_>
+    where
+        Reader: tokio::io::AsyncRead + Unpin,
+    {
+        // We return a Future that's sync (has a `wait` method) if and only if the provided stream is SyncProof.
+        // SyncFuture contract: we are only allowed to await on sync-proof futures, the AsyncRead and
+        // AsyncReadExt methods of the stream.
+        SyncFuture::new(async move {
+            // Each libpq message begins with a message type byte, followed by message length
+            // If the client closes the connection, return None. But if the client closes the
+            // connection in the middle of a message, we will return an error.
+            let tag = match retry_read!(stream.read_u8().await) {
+                Ok(b) => b,
+                Err(e) if e.kind() == io::ErrorKind::UnexpectedEof => return Ok(None),
+                Err(e) => return Err(ConnectionError::Socket(e)),
+            };
 
-        // length field includes itself, but not message type.
-        let total_len = len as usize + 1;
-        if buf.len() < total_len {
-            // Don't have full message yet.
-            let to_read = total_len - buf.len();
-            buf.reserve(to_read);
-            return Ok(None);
-        }
+            // The message length includes itself, so it better be at least 4.
+            let len = retry_read!(stream.read_u32().await)
+                .map_err(ConnectionError::Socket)?
+                .checked_sub(4)
+                .ok_or_else(|| ConnectionError::Protocol("invalid message length".to_string()))?;
 
-        // got the message, advance buffer
-        let mut msg = buf.split_to(total_len).freeze();
-        msg.advance(5); // consume message type and len
+            let body = {
+                let mut buffer = vec![0u8; len as usize];
+                stream
+                    .read_exact(&mut buffer)
+                    .await
+                    .map_err(ConnectionError::Socket)?;
+                Bytes::from(buffer)
+            };
 
-        match tag {
-            b'Q' => Ok(Some(FeMessage::Query(msg))),
-            b'P' => Ok(Some(FeParseMessage::parse(msg)?)),
-            b'D' => Ok(Some(FeDescribeMessage::parse(msg)?)),
-            b'E' => Ok(Some(FeExecuteMessage::parse(msg)?)),
-            b'B' => Ok(Some(FeBindMessage::parse(msg)?)),
-            b'C' => Ok(Some(FeCloseMessage::parse(msg)?)),
-            b'S' => Ok(Some(FeMessage::Sync)),
-            b'X' => Ok(Some(FeMessage::Terminate)),
-            b'd' => Ok(Some(FeMessage::CopyData(msg))),
-            b'c' => Ok(Some(FeMessage::CopyDone)),
-            b'f' => Ok(Some(FeMessage::CopyFail)),
-            b'p' => Ok(Some(FeMessage::PasswordMessage(msg))),
-            tag => Err(ProtocolError::Protocol(format!(
-                "unknown message tag: {tag},'{msg:?}'"
-            ))),
-        }
+            match tag {
+                b'Q' => Ok(Some(FeMessage::Query(body))),
+                b'P' => Ok(Some(FeParseMessage::parse(body)?)),
+                b'D' => Ok(Some(FeDescribeMessage::parse(body)?)),
+                b'E' => Ok(Some(FeExecuteMessage::parse(body)?)),
+                b'B' => Ok(Some(FeBindMessage::parse(body)?)),
+                b'C' => Ok(Some(FeCloseMessage::parse(body)?)),
+                b'S' => Ok(Some(FeMessage::Sync)),
+                b'X' => Ok(Some(FeMessage::Terminate)),
+                b'd' => Ok(Some(FeMessage::CopyData(body))),
+                b'c' => Ok(Some(FeMessage::CopyDone)),
+                b'f' => Ok(Some(FeMessage::CopyFail)),
+                b'p' => Ok(Some(FeMessage::PasswordMessage(body))),
+                tag => {
+                    return Err(ConnectionError::Protocol(format!(
+                        "unknown message tag: {tag},'{body:?}'"
+                    )))
+                }
+            }
+        })
     }
 }
 
 impl FeStartupPacket {
-    /// Read and parse startup message from the `buf` input buffer. It is
-    /// different from [`FeMessage::parse`] because startup messages don't have
-    /// message type byte; otherwise, its comments apply.
-    pub fn parse(buf: &mut BytesMut) -> Result<Option<FeStartupPacket>, ProtocolError> {
+    /// Read startup message from the stream.
+    // XXX: It's tempting yet undesirable to accept `stream` by value,
+    // since such a change will cause user-supplied &mut references to be consumed
+    pub fn read(
+        stream: &mut (impl io::Read + Unpin),
+    ) -> Result<Option<FeMessage>, ConnectionError> {
+        Self::read_fut(&mut AsyncishRead(stream)).wait()
+    }
+
+    /// Read startup message from the stream.
+    // XXX: It's tempting yet undesirable to accept `stream` by value,
+    // since such a change will cause user-supplied &mut references to be consumed
+    pub fn read_fut<Reader>(
+        stream: &mut Reader,
+    ) -> SyncFuture<Reader, impl Future<Output = Result<Option<FeMessage>, ConnectionError>> + '_>
+    where
+        Reader: tokio::io::AsyncRead + Unpin,
+    {
         const MAX_STARTUP_PACKET_LENGTH: usize = 10000;
         const RESERVED_INVALID_MAJOR_VERSION: u32 = 1234;
         const CANCEL_REQUEST_CODE: u32 = 5678;
         const NEGOTIATE_SSL_CODE: u32 = 5679;
         const NEGOTIATE_GSS_CODE: u32 = 5680;
 
-        // need at least 4 bytes with packet len
-        if buf.len() < 4 {
-            let to_read = 4 - buf.len();
-            buf.reserve(to_read);
-            return Ok(None);
-        }
+        SyncFuture::new(async move {
+            // Read length. If the connection is closed before reading anything (or before
+            // reading 4 bytes, to be precise), return None to indicate that the connection
+            // was closed. This matches the PostgreSQL server's behavior, which avoids noise
+            // in the log if the client opens connection but closes it immediately.
+            let len = match retry_read!(stream.read_u32().await) {
+                Ok(len) => len as usize,
+                Err(e) if e.kind() == io::ErrorKind::UnexpectedEof => return Ok(None),
+                Err(e) => return Err(ConnectionError::Socket(e)),
+            };
 
-        // We shouldn't advance `buf` as probably full message is not there yet,
-        // so can't directly use Bytes::get_u32 etc.
-        let len = (&buf[0..4]).read_u32::<BigEndian>().unwrap() as usize;
-        if len < 4 || len > MAX_STARTUP_PACKET_LENGTH {
-            return Err(ProtocolError::Protocol(format!(
-                "invalid startup packet message length {}",
-                len
-            )));
-        }
-
-        if buf.len() < len {
-            // Don't have full message yet.
-            let to_read = len - buf.len();
-            buf.reserve(to_read);
-            return Ok(None);
-        }
-
-        // got the message, advance buffer
-        let mut msg = buf.split_to(len).freeze();
-        msg.advance(4); // consume len
-
-        let request_code = msg.get_u32();
-        let req_hi = request_code >> 16;
-        let req_lo = request_code & ((1 << 16) - 1);
-        // StartupMessage, CancelRequest, SSLRequest etc are differentiated by request code.
-        let message = match (req_hi, req_lo) {
-            (RESERVED_INVALID_MAJOR_VERSION, CANCEL_REQUEST_CODE) => {
-                if msg.remaining() != 8 {
-                    return Err(ProtocolError::BadMessage(
-                        "CancelRequest message is malformed, backend PID / secret key missing"
-                            .to_owned(),
-                    ));
-                }
-                FeStartupPacket::CancelRequest(CancelKeyData {
-                    backend_pid: msg.get_i32(),
-                    cancel_key: msg.get_i32(),
-                })
-            }
-            (RESERVED_INVALID_MAJOR_VERSION, NEGOTIATE_SSL_CODE) => {
-                // Requested upgrade to SSL (aka TLS)
-                FeStartupPacket::SslRequest
-            }
-            (RESERVED_INVALID_MAJOR_VERSION, NEGOTIATE_GSS_CODE) => {
-                // Requested upgrade to GSSAPI
-                FeStartupPacket::GssEncRequest
-            }
-            (RESERVED_INVALID_MAJOR_VERSION, unrecognized_code) => {
-                return Err(ProtocolError::Protocol(format!(
-                    "Unrecognized request code {unrecognized_code}"
+            #[allow(clippy::manual_range_contains)]
+            if len < 4 || len > MAX_STARTUP_PACKET_LENGTH {
+                return Err(ConnectionError::Protocol(format!(
+                    "invalid message length {len}"
                 )));
             }
-            // TODO bail if protocol major_version is not 3?
-            (major_version, minor_version) => {
-                // StartupMessage
 
-                // Parse pairs of null-terminated strings (key, value).
-                // See `postgres: ProcessStartupPacket, build_startup_packet`.
-                let mut tokens = str::from_utf8(&msg)
-                    .map_err(|_e| {
-                        ProtocolError::BadMessage("StartupMessage params: invalid utf-8".to_owned())
-                    })?
-                    .strip_suffix('\0') // drop packet's own null
-                    .ok_or_else(|| {
-                        ProtocolError::Protocol(
-                            "StartupMessage params: missing null terminator".to_string(),
-                        )
-                    })?
-                    .split_terminator('\0');
+            let request_code =
+                retry_read!(stream.read_u32().await).map_err(ConnectionError::Socket)?;
 
-                let mut params = HashMap::new();
-                while let Some(name) = tokens.next() {
-                    let value = tokens.next().ok_or_else(|| {
-                        ProtocolError::Protocol(
-                            "StartupMessage params: key without value".to_string(),
-                        )
-                    })?;
+            // the rest of startup packet are params
+            let params_len = len - 8;
+            let mut params_bytes = vec![0u8; params_len];
+            stream
+                .read_exact(params_bytes.as_mut())
+                .await
+                .map_err(ConnectionError::Socket)?;
 
-                    params.insert(name.to_owned(), value.to_owned());
+            // Parse params depending on request code
+            let req_hi = request_code >> 16;
+            let req_lo = request_code & ((1 << 16) - 1);
+            let message = match (req_hi, req_lo) {
+                (RESERVED_INVALID_MAJOR_VERSION, CANCEL_REQUEST_CODE) => {
+                    if params_len != 8 {
+                        return Err(ConnectionError::Protocol(
+                            "expected 8 bytes for CancelRequest params".to_string(),
+                        ));
+                    }
+                    let mut cursor = Cursor::new(params_bytes);
+                    FeStartupPacket::CancelRequest(CancelKeyData {
+                        backend_pid: cursor.read_i32().await.map_err(ConnectionError::Socket)?,
+                        cancel_key: cursor.read_i32().await.map_err(ConnectionError::Socket)?,
+                    })
                 }
-
-                FeStartupPacket::StartupMessage {
-                    major_version,
-                    minor_version,
-                    params: StartupMessageParams { params },
+                (RESERVED_INVALID_MAJOR_VERSION, NEGOTIATE_SSL_CODE) => {
+                    // Requested upgrade to SSL (aka TLS)
+                    FeStartupPacket::SslRequest
                 }
-            }
-        };
-        Ok(Some(message))
+                (RESERVED_INVALID_MAJOR_VERSION, NEGOTIATE_GSS_CODE) => {
+                    // Requested upgrade to GSSAPI
+                    FeStartupPacket::GssEncRequest
+                }
+                (RESERVED_INVALID_MAJOR_VERSION, unrecognized_code) => {
+                    return Err(ConnectionError::Protocol(format!(
+                        "Unrecognized request code {unrecognized_code}"
+                    )));
+                }
+                // TODO bail if protocol major_version is not 3?
+                (major_version, minor_version) => {
+                    // Parse pairs of null-terminated strings (key, value).
+                    // See `postgres: ProcessStartupPacket, build_startup_packet`.
+                    let mut tokens = str::from_utf8(&params_bytes)
+                        .context("StartupMessage params: invalid utf-8")?
+                        .strip_suffix('\0') // drop packet's own null
+                        .ok_or_else(|| {
+                            ConnectionError::Protocol(
+                                "StartupMessage params: missing null terminator".to_string(),
+                            )
+                        })?
+                        .split_terminator('\0');
+
+                    let mut params = HashMap::new();
+                    while let Some(name) = tokens.next() {
+                        let value = tokens.next().ok_or_else(|| {
+                            ConnectionError::Protocol(
+                                "StartupMessage params: key without value".to_string(),
+                            )
+                        })?;
+
+                        params.insert(name.to_owned(), value.to_owned());
+                    }
+
+                    FeStartupPacket::StartupMessage {
+                        major_version,
+                        minor_version,
+                        params: StartupMessageParams { params },
+                    }
+                }
+            };
+
+            Ok(Some(FeMessage::StartupPacket(message)))
+        })
     }
 }
 
 impl FeParseMessage {
-    fn parse(mut buf: Bytes) -> Result<FeMessage, ProtocolError> {
+    fn parse(mut buf: Bytes) -> anyhow::Result<FeMessage> {
         // FIXME: the rust-postgres driver uses a named prepared statement
         // for copy_out(). We're not prepared to handle that correctly. For
         // now, just ignore the statement name, assuming that the client never
@@ -390,82 +440,55 @@ impl FeParseMessage {
 
         let _pstmt_name = read_cstr(&mut buf)?;
         let query_string = read_cstr(&mut buf)?;
-        if buf.remaining() < 2 {
-            return Err(ProtocolError::BadMessage(
-                "Parse message is malformed, nparams missing".to_string(),
-            ));
-        }
         let nparams = buf.get_i16();
 
-        if nparams != 0 {
-            return Err(ProtocolError::BadMessage(
-                "query params not implemented".to_string(),
-            ));
-        }
+        ensure!(nparams == 0, "query params not implemented");
 
         Ok(FeMessage::Parse(FeParseMessage { query_string }))
     }
 }
 
 impl FeDescribeMessage {
-    fn parse(mut buf: Bytes) -> Result<FeMessage, ProtocolError> {
+    fn parse(mut buf: Bytes) -> anyhow::Result<FeMessage> {
         let kind = buf.get_u8();
         let _pstmt_name = read_cstr(&mut buf)?;
 
         // FIXME: see FeParseMessage::parse
-        if kind != b'S' {
-            return Err(ProtocolError::BadMessage(
-                "only prepared statemement Describe is implemented".to_string(),
-            ));
-        }
+        ensure!(
+            kind == b'S',
+            "only prepared statemement Describe is implemented"
+        );
 
         Ok(FeMessage::Describe(FeDescribeMessage { kind }))
     }
 }
 
 impl FeExecuteMessage {
-    fn parse(mut buf: Bytes) -> Result<FeMessage, ProtocolError> {
+    fn parse(mut buf: Bytes) -> anyhow::Result<FeMessage> {
         let portal_name = read_cstr(&mut buf)?;
-        if buf.remaining() < 4 {
-            return Err(ProtocolError::BadMessage(
-                "FeExecuteMessage message is malformed, maxrows missing".to_string(),
-            ));
-        }
         let maxrows = buf.get_i32();
 
-        if !portal_name.is_empty() {
-            return Err(ProtocolError::BadMessage(
-                "named portals not implemented".to_string(),
-            ));
-        }
-        if maxrows != 0 {
-            return Err(ProtocolError::BadMessage(
-                "row limit in Execute message not implemented".to_string(),
-            ));
-        }
+        ensure!(portal_name.is_empty(), "named portals not implemented");
+        ensure!(maxrows == 0, "row limit in Execute message not implemented");
 
         Ok(FeMessage::Execute(FeExecuteMessage { maxrows }))
     }
 }
 
 impl FeBindMessage {
-    fn parse(mut buf: Bytes) -> Result<FeMessage, ProtocolError> {
+    fn parse(mut buf: Bytes) -> anyhow::Result<FeMessage> {
         let portal_name = read_cstr(&mut buf)?;
         let _pstmt_name = read_cstr(&mut buf)?;
 
         // FIXME: see FeParseMessage::parse
-        if !portal_name.is_empty() {
-            return Err(ProtocolError::BadMessage(
-                "named portals not implemented".to_string(),
-            ));
-        }
+        ensure!(portal_name.is_empty(), "named portals not implemented");
 
         Ok(FeMessage::Bind(FeBindMessage))
     }
 }
 
 impl FeCloseMessage {
-    fn parse(mut buf: Bytes) -> Result<FeMessage, ProtocolError> {
+    fn parse(mut buf: Bytes) -> anyhow::Result<FeMessage> {
         let _kind = buf.get_u8();
         let _pstmt_or_portal_name = read_cstr(&mut buf)?;
 
@@ -494,7 +517,6 @@ pub enum BeMessage<'a> {
     CloseComplete,
     // None means column is NULL
     DataRow(&'a [Option<&'a [u8]>]),
-    // None errcode means internal_error will be sent.
     ErrorResponse(&'a str, Option<&'a [u8; 5]>),
     /// Single byte - used in response to SSLRequest/GSSENCRequest.
     EncryptionResponse(bool),
@@ -525,11 +547,6 @@ impl<'a> BeMessage<'a> {
         value: b"UTF8",
     };
 
-    pub const INTEGER_DATETIMES: Self = Self::ParameterStatus {
-        name: b"integer_datetimes",
-        value: b"on",
-    };
-
     /// Build a [`BeMessage::ParameterStatus`] holding the server version.
     pub fn server_version(version: &'a str) -> Self {
         Self::ParameterStatus {
@@ -608,7 +625,7 @@ impl RowDescriptor<'_> {
 #[derive(Debug)]
 pub struct XLogDataBody<'a> {
     pub wal_start: u64,
-    pub wal_end: u64, // current end of WAL on the server
+    pub wal_end: u64,
     pub timestamp: i64,
     pub data: &'a [u8],
 }
@@ -648,11 +665,12 @@ fn write_body<R>(buf: &mut BytesMut, f: impl FnOnce(&mut BytesMut) -> R) -> R {
 }
 
 /// Safe write of s into buf as cstring (String in the protocol).
-fn write_cstr(s: impl AsRef<[u8]>, buf: &mut BytesMut) -> Result<(), ProtocolError> {
+fn write_cstr(s: impl AsRef<[u8]>, buf: &mut BytesMut) -> io::Result<()> {
     let bytes = s.as_ref();
     if bytes.contains(&0) {
-        return Err(ProtocolError::BadMessage(
-            "string contains embedded null".to_owned(),
+        return Err(io::Error::new(
+            io::ErrorKind::InvalidInput,
+            "string contains embedded null",
         ));
     }
     buf.put_slice(bytes);
@@ -660,27 +678,22 @@ fn write_cstr(s: impl AsRef<[u8]>, buf: &mut BytesMut) -> Result<(), ProtocolErr
     Ok(())
 }
 
-/// Read cstring from buf, advancing it.
-fn read_cstr(buf: &mut Bytes) -> Result<Bytes, ProtocolError> {
-    let pos = buf
-        .iter()
-        .position(|x| *x == 0)
-        .ok_or_else(|| ProtocolError::BadMessage("missing cstring terminator".to_owned()))?;
-    let result = buf.split_to(pos);
+fn read_cstr(buf: &mut Bytes) -> anyhow::Result<Bytes> {
+    let pos = buf.iter().position(|x| *x == 0);
+    let result = buf.split_to(pos.context("missing terminator")?);
     buf.advance(1); // drop the null terminator
     Ok(result)
 }
 
 pub const SQLSTATE_INTERNAL_ERROR: &[u8; 5] = b"XX000";
-pub const SQLSTATE_SUCCESSFUL_COMPLETION: &[u8; 5] = b"00000";
 
 impl<'a> BeMessage<'a> {
-    /// Serialize `message` to the given `buf`.
-    /// Apart from smart memory managemet, BytesMut is good here as msg len
-    /// precedes its body and it is handy to write it down first and then fill
-    /// the length. With Write we would have to either calc it manually or have
-    /// one more buffer.
-    pub fn write(buf: &mut BytesMut, message: &BeMessage) -> Result<(), ProtocolError> {
+    /// Write message to the given buf.
+    // Unlike the reading side, we use BytesMut
+    // here as msg len precedes its body and it is handy to write it down first
+    // and then fill the length. With Write we would have to either calc it
+    // manually or have one more buffer.
+    pub fn write(buf: &mut BytesMut, message: &BeMessage) -> io::Result<()> {
         match message {
             BeMessage::AuthenticationOk => {
                 buf.put_u8(b'R');
@@ -725,7 +738,7 @@ impl<'a> BeMessage<'a> {
                             buf.put_slice(extra);
                         }
                     }
-                    Ok(())
+                    Ok::<_, io::Error>(())
                 })?;
             }
 
@@ -829,7 +842,7 @@ impl<'a> BeMessage<'a> {
                     write_cstr(error_msg, buf)?;
 
                     buf.put_u8(0); // terminator
-                    Ok(())
+                    Ok::<_, io::Error>(())
                 })?;
             }
 
@@ -852,7 +865,7 @@ impl<'a> BeMessage<'a> {
                     write_cstr(error_msg.as_bytes(), buf)?;
 
                     buf.put_u8(0); // terminator
-                    Ok(())
+                    Ok::<_, io::Error>(())
                 })?;
             }
 
@@ -907,7 +920,7 @@ impl<'a> BeMessage<'a> {
                         buf.put_i32(-1); /* typmod */
                         buf.put_i16(0); /* format code */
                     }
-                    Ok(())
+                    Ok::<_, io::Error>(())
                 })?;
             }
 
@@ -974,7 +987,7 @@ impl ReplicationFeedback {
     // null-terminated string - key,
     // uint32 - value length in bytes
     // value itself
-    pub fn serialize(&self, buf: &mut BytesMut) {
+    pub fn serialize(&self, buf: &mut BytesMut) -> Result<()> {
         buf.put_u8(REPLICATION_FEEDBACK_FIELDS_NUMBER); // # of keys
         buf.put_slice(b"current_timeline_size\0");
         buf.put_i32(8);
@@ -999,6 +1012,7 @@ impl ReplicationFeedback {
         buf.put_slice(b"ps_replytime\0");
         buf.put_i32(8);
         buf.put_i64(timestamp);
+        Ok(())
     }
 
     // Deserialize ReplicationFeedback message
@@ -1066,7 +1080,7 @@ mod tests {
         // because it is rounded up to microseconds during serialization.
         rf.ps_replytime = *PG_EPOCH + Duration::from_secs(100_000_000);
         let mut data = BytesMut::new();
-        rf.serialize(&mut data);
+        rf.serialize(&mut data).unwrap();
 
         let rf_parsed = ReplicationFeedback::parse(data.freeze());
         assert_eq!(rf, rf_parsed);
@@ -1081,7 +1095,7 @@ mod tests {
         // because it is rounded up to microseconds during serialization.
         rf.ps_replytime = *PG_EPOCH + Duration::from_secs(100_000_000);
         let mut data = BytesMut::new();
-        rf.serialize(&mut data);
+        rf.serialize(&mut data).unwrap();
 
         // Add an extra field to the buffer and adjust number of keys
         if let Some(first) = data.first_mut() {
@@ -1123,6 +1137,15 @@ mod tests {
         let params = make_params("foo\\ bar \\ \\\\ baz\\  lol");
         assert_eq!(split_options(&params), ["foo bar", " \\", "baz ", "lol"]);
     }
+
+    // Make sure that `read` is sync/async callable
+    async fn _assert(stream: &mut (impl tokio::io::AsyncRead + Unpin)) {
+        let _ = FeMessage::read(&mut [].as_ref());
+        let _ = FeMessage::read_fut(stream).await;
+
+        let _ = FeStartupPacket::read(&mut [].as_ref());
+        let _ = FeStartupPacket::read_fut(stream).await;
+    }
 }
 
 fn terminate_code(code: &[u8; 5]) -> [u8; 6] {

libs/pq_proto/src/sync.rs

@@ -0,0 +1,179 @@
+use pin_project_lite::pin_project;
+use std::future::Future;
+use std::marker::PhantomData;
+use std::pin::Pin;
+use std::{io, task};
+
+pin_project! {
+    /// We use this future to mark certain methods
+    /// as callable in both sync and async modes.
+    #[repr(transparent)]
+    pub struct SyncFuture<S, T: Future> {
+        #[pin]
+        inner: T,
+        _marker: PhantomData<S>,
+    }
+}
+
+/// This wrapper lets us synchronously wait for inner future's completion
+/// (see [`SyncFuture::wait`]) **provided that `S` implements [`SyncProof`]**.
+/// For instance, `S` may be substituted with types implementing
+/// [`tokio::io::AsyncRead`], but it's not the only viable option.
+impl<S, T: Future> SyncFuture<S, T> {
+    /// NOTE: caller should carefully pick a type for `S`,
+    /// because we don't want to enable [`SyncFuture::wait`] when
+    /// it's in fact impossible to run the future synchronously.
+    /// Violation of this contract will not cause UB, but
+    /// panics and async event loop freezes won't please you.
+    ///
+    /// Example:
+    ///
+    /// ```
+    /// # use pq_proto::sync::SyncFuture;
+    /// # use std::future::Future;
+    /// # use tokio::io::AsyncReadExt;
+    /// #
+    /// // Parse a pair of numbers from a stream
+    /// pub fn parse_pair<Reader>(
+    ///     stream: &mut Reader,
+    /// ) -> SyncFuture<Reader, impl Future<Output = anyhow::Result<(u32, u64)>> + '_>
+    /// where
+    ///     Reader: tokio::io::AsyncRead + Unpin,
+    /// {
+    ///     // If `Reader` is a `SyncProof`, this will give caller
+    ///     // an opportunity to use `SyncFuture::wait`, because
+    ///     // `.await` will always result in `Poll::Ready`.
+    ///     SyncFuture::new(async move {
+    ///         let x = stream.read_u32().await?;
+    ///         let y = stream.read_u64().await?;
+    ///         Ok((x, y))
+    ///     })
+    /// }
+    /// ```
+    pub fn new(inner: T) -> Self {
+        Self {
+            inner,
+            _marker: PhantomData,
+        }
+    }
+}
+
+impl<S, T: Future> Future for SyncFuture<S, T> {
+    type Output = T::Output;
+
+    /// In async code, [`SyncFuture`] behaves like a regular wrapper.
+    #[inline(always)]
+    fn poll(self: Pin<&mut Self>, cx: &mut task::Context<'_>) -> task::Poll<Self::Output> {
+        self.project().inner.poll(cx)
+    }
+}
+
+/// Postulates that we can call [`SyncFuture::wait`].
+/// If implementer is also a [`Future`], it should always
+/// return [`task::Poll::Ready`] from [`Future::poll`].
+///
+/// Each implementation should document which futures
+/// specifically are being declared sync-proof.
+pub trait SyncPostulate {}
+
+impl<T: SyncPostulate> SyncPostulate for &T {}
+impl<T: SyncPostulate> SyncPostulate for &mut T {}
+
+impl<P: SyncPostulate, T: Future> SyncFuture<P, T> {
+    /// Synchronously wait for future completion.
+    pub fn wait(mut self) -> T::Output {
+        const RAW_WAKER: task::RawWaker = task::RawWaker::new(
+            std::ptr::null(),
+            &task::RawWakerVTable::new(
+                |_| RAW_WAKER,
+                |_| panic!("SyncFuture: failed to wake"),
+                |_| panic!("SyncFuture: failed to wake by ref"),
+                |_| { /* drop is no-op */ },
+            ),
+        );
+
+        // SAFETY: We never move `self` during this call;
+        // furthermore, it will be dropped in the end regardless of panics
+        let this = unsafe { Pin::new_unchecked(&mut self) };
+
+        // SAFETY: This waker doesn't do anything apart from panicking
+        let waker = unsafe { task::Waker::from_raw(RAW_WAKER) };
+        let context = &mut task::Context::from_waker(&waker);
+
+        match this.poll(context) {
+            task::Poll::Ready(res) => res,
+            _ => panic!("SyncFuture: unexpected pending!"),
+        }
+    }
+}
+
+/// This wrapper turns any [`std::io::Read`] into a blocking [`tokio::io::AsyncRead`],
+/// which lets us abstract over sync & async readers in methods returning [`SyncFuture`].
+/// NOTE: you **should not** use this in async code.
+#[repr(transparent)]
+pub struct AsyncishRead<T: io::Read + Unpin>(pub T);
+
+/// This lets us call [`SyncFuture<AsyncishRead<_>, _>::wait`],
+/// and allows the future to await on any of the [`AsyncRead`]
+/// and [`AsyncReadExt`] methods on `AsyncishRead`.
+impl<T: io::Read + Unpin> SyncPostulate for AsyncishRead<T> {}
+
+impl<T: io::Read + Unpin> tokio::io::AsyncRead for AsyncishRead<T> {
+    #[inline(always)]
+    fn poll_read(
+        mut self: Pin<&mut Self>,
+        _cx: &mut task::Context<'_>,
+        buf: &mut tokio::io::ReadBuf<'_>,
+    ) -> task::Poll<io::Result<()>> {
+        task::Poll::Ready(
+            // `Read::read` will block, meaning we don't need a real event loop!
+            self.0
+                .read(buf.initialize_unfilled())
+                .map(|sz| buf.advance(sz)),
+        )
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use tokio::io::{AsyncReadExt, AsyncWriteExt};
+
+    // async helper(stream: &mut impl AsyncRead) -> io::Result<u32>
+    fn bytes_add<Reader>(
+        stream: &mut Reader,
+    ) -> SyncFuture<Reader, impl Future<Output = io::Result<u32>> + '_>
+    where
+        Reader: tokio::io::AsyncRead + Unpin,
+    {
+        SyncFuture::new(async move {
+            let a = stream.read_u32().await?;
+            let b = stream.read_u32().await?;
+            Ok(a + b)
+        })
+    }
+
+    #[test]
+    fn test_sync() {
+        let bytes = [100u32.to_be_bytes(), 200u32.to_be_bytes()].concat();
+        let res = bytes_add(&mut AsyncishRead(&mut &bytes[..]))
+            .wait()
+            .unwrap();
+        assert_eq!(res, 300);
+    }
+
+    // We need a single-threaded executor for this test
+    #[tokio::test(flavor = "current_thread")]
+    async fn test_async() {
+        let (mut tx, mut rx) = tokio::net::UnixStream::pair().unwrap();
+
+        let write = async move {
+            tx.write_u32(100).await?;
+            tx.write_u32(200).await?;
+            Ok(())
+        };
+
+        let (res, ()) = tokio::try_join!(bytes_add(&mut rx), write).unwrap();
+        assert_eq!(res, 300);
+    }
+}

libs/remote_storage/Cargo.toml

@@ -21,7 +21,7 @@ toml_edit.workspace = true
 tracing.workspace = true
 metrics.workspace = true
 utils.workspace = true
-pin-project-lite.workspace = true
+
 workspace_hack.workspace = true
 
 [dev-dependencies]

libs/remote_storage/src/lib.rs

@@ -111,7 +111,7 @@ pub trait RemoteStorage: Send + Sync + 'static {
 }
 
 pub struct Download {
-    pub download_stream: Pin<Box<dyn io::AsyncRead + Unpin + Send + Sync>>,
+    pub download_stream: Pin<Box<dyn io::AsyncRead + Unpin + Send>>,
     /// Extra key-value data, associated with the current remote file.
     pub metadata: Option<StorageMetadata>,
 }

libs/remote_storage/src/s3_bucket.rs

@@ -20,10 +20,7 @@ use aws_sdk_s3::{
 };
 use aws_smithy_http::body::SdkBody;
 use hyper::Body;
-use tokio::{
-    io::{self, AsyncRead},
-    sync::Semaphore,
-};
+use tokio::{io, sync::Semaphore};
 use tokio_util::io::ReaderStream;
 use tracing::debug;
 
@@ -105,7 +102,7 @@ pub struct S3Bucket {
     // Every request to S3 can be throttled or cancelled, if a certain number of requests per second is exceeded.
     // Same goes to IAM, which is queried before every S3 request, if enabled. IAM has even lower RPS threshold.
     // The helps to ensure we don't exceed the thresholds.
-    concurrency_limiter: Arc<Semaphore>,
+    concurrency_limiter: Semaphore,
 }
 
 #[derive(Default)]
@@ -165,7 +162,7 @@ impl S3Bucket {
             client,
             bucket_name: aws_config.bucket_name.clone(),
             prefix_in_bucket,
-            concurrency_limiter: Arc::new(Semaphore::new(aws_config.concurrency_limit.get())),
+            concurrency_limiter: Semaphore::new(aws_config.concurrency_limit.get()),
         })
     }
 
@@ -197,10 +194,9 @@ impl S3Bucket {
     }
 
     async fn download_object(&self, request: GetObjectRequest) -> Result<Download, DownloadError> {
-        let permit = self
+        let _guard = self
             .concurrency_limiter
-            .clone()
-            .acquire_owned()
+            .acquire()
             .await
             .context("Concurrency limiter semaphore got closed during S3 download")
             .map_err(DownloadError::Other)?;
@@ -221,10 +217,9 @@ impl S3Bucket {
                 let metadata = object_output.metadata().cloned().map(StorageMetadata);
                 Ok(Download {
                     metadata,
-                    download_stream: Box::pin(io::BufReader::new(RatelimitedAsyncRead::new(
-                        permit,
+                    download_stream: Box::pin(io::BufReader::new(
                         object_output.body.into_async_read(),
-                    ))),
+                    )),
                 })
             }
             Err(SdkError::ServiceError {
@@ -245,32 +240,6 @@ impl S3Bucket {
     }
 }
 
-pin_project_lite::pin_project! {
-    /// An `AsyncRead` adapter which carries a permit for the lifetime of the value.
-    struct RatelimitedAsyncRead<S> {
-        permit: tokio::sync::OwnedSemaphorePermit,
-        #[pin]
-        inner: S,
-    }
-}
-
-impl<S: AsyncRead> RatelimitedAsyncRead<S> {
-    fn new(permit: tokio::sync::OwnedSemaphorePermit, inner: S) -> Self {
-        RatelimitedAsyncRead { permit, inner }
-    }
-}
-
-impl<S: AsyncRead> AsyncRead for RatelimitedAsyncRead<S> {
-    fn poll_read(
-        self: std::pin::Pin<&mut Self>,
-        cx: &mut std::task::Context<'_>,
-        buf: &mut io::ReadBuf<'_>,
-    ) -> std::task::Poll<std::io::Result<()>> {
-        let this = self.project();
-        this.inner.poll_read(cx, buf)
-    }
-}
-
 #[async_trait::async_trait]
 impl RemoteStorage for S3Bucket {
     async fn list(&self) -> anyhow::Result<Vec<RemotePath>> {

libs/tenant_size_model/Cargo.toml

@@ -7,7 +7,5 @@ license.workspace = true
 
 [dependencies]
 anyhow.workspace = true
-serde.workspace = true
-serde_json.workspace = true
 
 workspace_hack.workspace = true

libs/tenant_size_model/src/calculation.rs

@@ -1,219 +0,0 @@
-use crate::{SegmentMethod, SegmentSizeResult, SizeResult, StorageModel};
-
-//
-//                 *-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.
-#[derive(Clone, Debug)]
-struct SegmentSize {
-    method: SegmentMethod,
-
-    // calculated size of this subtree, using this method
-    accum_size: u64,
-
-    seg_id: usize,
-    children: Vec<SegmentSize>,
-}
-
-struct SizeAlternatives {
-    // cheapest alternative if parent is available.
-    incremental: SegmentSize,
-
-    // cheapest alternative if parent node is not available
-    non_incremental: Option<SegmentSize>,
-}
-
-impl StorageModel {
-    pub fn calculate(&self) -> SizeResult {
-        // Build adjacency list. 'child_list' is indexed by segment id. Each entry
-        // contains a list of all child segments of the segment.
-        let mut roots: Vec<usize> = Vec::new();
-        let mut child_list: Vec<Vec<usize>> = Vec::new();
-        child_list.resize(self.segments.len(), Vec::new());
-
-        for (seg_id, seg) in self.segments.iter().enumerate() {
-            if let Some(parent_id) = seg.parent {
-                child_list[parent_id].push(seg_id);
-            } else {
-                roots.push(seg_id);
-            }
-        }
-
-        let mut segment_results = Vec::new();
-        segment_results.resize(
-            self.segments.len(),
-            SegmentSizeResult {
-                method: SegmentMethod::Skipped,
-                accum_size: 0,
-            },
-        );
-
-        let mut total_size = 0;
-        for root in roots {
-            if let Some(selected) = self.size_here(root, &child_list).non_incremental {
-                StorageModel::fill_selected_sizes(&selected, &mut segment_results);
-                total_size += selected.accum_size;
-            } else {
-                // Couldn't find any way to get this root. Error?
-            }
-        }
-
-        SizeResult {
-            total_size,
-            segments: segment_results,
-        }
-    }
-
-    fn fill_selected_sizes(selected: &SegmentSize, result: &mut Vec<SegmentSizeResult>) {
-        result[selected.seg_id] = SegmentSizeResult {
-            method: selected.method,
-            accum_size: selected.accum_size,
-        };
-        // recurse to children
-        for child in selected.children.iter() {
-            StorageModel::fill_selected_sizes(child, result);
-        }
-    }
-
-    //
-    // This is the core of the sizing calculation.
-    //
-    // This is a recursive function, that for each Segment calculates the best way
-    // to reach all the Segments that are marked as needed in this subtree, under two
-    // different conditions:
-    // a) when the parent of this segment is available (as a snaphot or through WAL), and
-    // b) when the parent of this segment is not available.
-    //
-    fn size_here(&self, seg_id: usize, child_list: &Vec<Vec<usize>>) -> SizeAlternatives {
-        let seg = &self.segments[seg_id];
-        // First figure out the best way to get each child
-        let mut children = Vec::new();
-        for child_id in &child_list[seg_id] {
-            children.push(self.size_here(*child_id, child_list))
-        }
-
-        // Method 1. If this node is not needed, we can skip it as long as we
-        // take snapshots later in each sub-tree
-        let snapshot_later = if !seg.needed {
-            let mut snapshot_later = SegmentSize {
-                seg_id,
-                method: SegmentMethod::Skipped,
-                accum_size: 0,
-                children: Vec::new(),
-            };
-
-            let mut possible = true;
-            for child in children.iter() {
-                if let Some(non_incremental) = &child.non_incremental {
-                    snapshot_later.accum_size += non_incremental.accum_size;
-                    snapshot_later.children.push(non_incremental.clone())
-                } else {
-                    possible = false;
-                    break;
-                }
-            }
-            if possible {
-                Some(snapshot_later)
-            } else {
-                None
-            }
-        } else {
-            None
-        };
-
-        // Method 2. Get a snapshot here. This assumed to be possible, if the 'size' of
-        // this Segment was given.
-        let snapshot_here = if !seg.needed || seg.parent.is_none() {
-            if let Some(snapshot_size) = seg.size {
-                let mut snapshot_here = SegmentSize {
-                    seg_id,
-                    method: SegmentMethod::SnapshotHere,
-                    accum_size: snapshot_size,
-                    children: Vec::new(),
-                };
-                for child in children.iter() {
-                    snapshot_here.accum_size += child.incremental.accum_size;
-                    snapshot_here.children.push(child.incremental.clone())
-                }
-                Some(snapshot_here)
-            } else {
-                None
-            }
-        } else {
-            None
-        };
-
-        // Method 3. Use WAL to get here from parent
-        let wal_here = {
-            let mut wal_here = SegmentSize {
-                seg_id,
-                method: SegmentMethod::Wal,
-                accum_size: if let Some(parent_id) = seg.parent {
-                    seg.lsn - self.segments[parent_id].lsn
-                } else {
-                    0
-                },
-                children: Vec::new(),
-            };
-            for child in children {
-                wal_here.accum_size += child.incremental.accum_size;
-                wal_here.children.push(child.incremental)
-            }
-            wal_here
-        };
-
-        // If the parent is not available, what's the cheapest method involving
-        // a snapshot here or later?
-        let mut cheapest_non_incremental: Option<SegmentSize> = None;
-        if let Some(snapshot_here) = snapshot_here {
-            cheapest_non_incremental = Some(snapshot_here);
-        }
-        if let Some(snapshot_later) = snapshot_later {
-            // Use <=, to prefer skipping if the size is equal
-            if let Some(parent) = &cheapest_non_incremental {
-                if snapshot_later.accum_size <= parent.accum_size {
-                    cheapest_non_incremental = Some(snapshot_later);
-                }
-            } else {
-                cheapest_non_incremental = Some(snapshot_later);
-            }
-        }
-
-        // And what's the cheapest method, if the parent is available?
-        let cheapest_incremental = if let Some(cheapest_non_incremental) = &cheapest_non_incremental
-        {
-            // Is it cheaper to use a snapshot here or later, anyway?
-            // Use <, to prefer Wal over snapshot if the cost is the same
-            if wal_here.accum_size < cheapest_non_incremental.accum_size {
-                wal_here
-            } else {
-                cheapest_non_incremental.clone()
-            }
-        } else {
-            wal_here
-        };
-
-        SizeAlternatives {
-            incremental: cheapest_incremental,
-            non_incremental: cheapest_non_incremental,
-        }
-    }
-}

libs/tenant_size_model/src/lib.rs

@@ -1,70 +1,401 @@
-//! Synthetic size calculation
+use std::borrow::Cow;
+use std::collections::HashMap;
 
-mod calculation;
-pub mod svg;
+use anyhow::Context;
 
-/// StorageModel is the input to the synthetic size calculation. It represents
-/// a tree of timelines, with just the information that's needed for the
-/// calculation. This doesn't track timeline names or where each timeline
-/// begins and ends, for example. Instead, it consists of "points of interest"
-/// on the timelines. A point of interest could be the timeline start or end point,
-/// the oldest point on a timeline that needs to be retained because of PITR
-/// cutoff, or snapshot points named by the user. For each such point, and the
-/// edge connecting the points (implicit in Segment), we store information about
-/// whether we need to be able to recover to the point, and if known, the logical
-/// size at the point.
+/// Pricing model or history size builder.
 ///
-/// The segments must form a well-formed tree, with no loops.
-#[derive(serde::Serialize)]
-pub struct StorageModel {
-    pub segments: Vec<Segment>,
+/// 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>,
 }
 
-/// Segment represents one point in the tree of branches, *and* the edge that leads
-/// to it (if any). We don't need separate structs for points and edges, because each
-/// point can have only one parent.
-///
-/// When 'needed' is true, it means that we need to be able to reconstruct
-/// any version between 'parent.lsn' and 'lsn'. If you want to represent that only
-/// a single point is needed, create two Segments with the same lsn, and mark only
-/// the child as needed.
-///
-#[derive(Clone, Debug, Eq, PartialEq, serde::Serialize, serde::Deserialize)]
+/// Snapshot of a branch.
+#[derive(Clone, Debug, Eq, PartialEq)]
 pub struct Segment {
     /// Previous segment index into ['Storage::segments`], if any.
-    pub parent: Option<usize>,
+    parent: Option<usize>,
 
-    /// LSN at this point
-    pub lsn: u64,
+    /// 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>,
 
-    /// Logical size at this node, if known.
-    pub size: Option<u64>,
+    /// LSN before this state
+    start_lsn: u64,
 
-    /// If true, the segment from parent to this node is needed by `retention_period`
+    /// LSN at this state
+    pub end_lsn: u64,
+
+    /// Logical size before this state
+    start_size: u64,
+
+    /// Logical size at this state. Can be None in the last Segment of a branch.
+    pub end_size: Option<u64>,
+
+    /// Indices to [`Storage::segments`]
+    ///
+    /// FIXME: this could be an Option<usize>
+    children_after: Vec<usize>,
+
+    /// Determined by `retention_period` given to [`Storage::calculate`]
     pub needed: bool,
 }
 
-/// Result of synthetic size calculation. Returned by StorageModel::calculate()
-pub struct SizeResult {
-    pub total_size: u64,
+//
+//
+//
+//
+//                 *-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
 
-    // This has same length as the StorageModel::segments vector in the input.
-    // Each entry in this array corresponds to the entry with same index in
-    // StorageModel::segments.
-    pub segments: Vec<SegmentSizeResult>,
+/// [`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>,
 }
 
-#[derive(Clone, Debug, Eq, PartialEq, serde::Serialize, serde::Deserialize)]
-pub struct SegmentSizeResult {
-    pub method: SegmentMethod,
-    // calculated size of this subtree, using this method
-    pub accum_size: u64,
+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, serde::Serialize, serde::Deserialize)]
+#[derive(Clone, Copy, Debug, Eq, PartialEq)]
 pub enum SegmentMethod {
-    SnapshotHere, // A logical snapshot is needed after this segment
-    Wal,          // Keep WAL leading up to this node
+    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: Some(0),
+            children_after: Vec::new(),
+        };
+
+        Storage {
+            segments: vec![init_segment],
+            branches: HashMap::from([(initial_branch, 0)]),
+        }
+    }
+
+    /// Advances the branch with a new point, at given LSN.
+    pub fn insert_point<Q: ?Sized>(
+        &mut self,
+        branch: &Q,
+        op: Cow<'static, str>,
+        lsn: u64,
+        size: Option<u64>,
+    ) -> anyhow::Result<()>
+    where
+        K: std::borrow::Borrow<Q>,
+        Q: std::hash::Hash + Eq + std::fmt::Debug,
+    {
+        let Some(lastseg_id) = self.branches.get(branch).copied() else { anyhow::bail!("branch not found: {branch:?}") };
+        let newseg_id = self.segments.len();
+        let lastseg = &mut self.segments[lastseg_id];
+
+        assert!(lsn > lastseg.end_lsn);
+
+        let Some(start_size) = lastseg.end_size else { anyhow::bail!("no end_size on latest segment for {branch:?}") };
+
+        let newseg = Segment {
+            op,
+            parent: Some(lastseg_id),
+            start_lsn: lastseg.end_lsn,
+            end_lsn: lsn,
+            start_size,
+            end_size: size,
+            children_after: Vec::new(),
+            needed: false,
+        };
+        lastseg.children_after.push(newseg_id);
+
+        self.segments.push(newseg);
+        *self.branches.get_mut(branch).expect("read already") = newseg_id;
+
+        Ok(())
+    }
+
+    /// 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,
+    ) -> anyhow::Result<()>
+    where
+        K: std::borrow::Borrow<Q>,
+        Q: std::hash::Hash + Eq + std::fmt::Debug,
+    {
+        let Some(lastseg_id) = self.branches.get(branch).copied() else { anyhow::bail!("branch not found: {branch:?}") };
+        let newseg_id = self.segments.len();
+        let lastseg = &mut self.segments[lastseg_id];
+
+        let Some(last_end_size) = lastseg.end_size else { anyhow::bail!("no end_size on latest segment for {branch:?}") };
+
+        let newseg = Segment {
+            op,
+            parent: Some(lastseg_id),
+            start_lsn: lastseg.end_lsn,
+            end_lsn: lastseg.end_lsn + lsn_bytes,
+            start_size: last_end_size,
+            end_size: Some((last_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;
+        Ok(())
+    }
+
+    pub fn insert<Q: ?Sized>(&mut self, branch: &Q, bytes: u64) -> anyhow::Result<()>
+    where
+        K: std::borrow::Borrow<Q>,
+        Q: std::hash::Hash + Eq + std::fmt::Debug,
+    {
+        self.modify_branch(branch, "insert".into(), bytes, bytes as i64)
+    }
+
+    pub fn update<Q: ?Sized>(&mut self, branch: &Q, bytes: u64) -> anyhow::Result<()>
+    where
+        K: std::borrow::Borrow<Q>,
+        Q: std::hash::Hash + Eq + std::fmt::Debug,
+    {
+        self.modify_branch(branch, "update".into(), bytes, 0i64)
+    }
+
+    pub fn delete<Q: ?Sized>(&mut self, branch: &Q, bytes: u64) -> anyhow::Result<()>
+    where
+        K: std::borrow::Borrow<Q>,
+        Q: std::hash::Hash + Eq + std::fmt::Debug,
+    {
+        self.modify_branch(branch, "delete".into(), bytes, -(bytes as i64))
+    }
+
+    pub fn branch<Q: ?Sized>(&mut self, parent: &Q, name: K) -> anyhow::Result<()>
+    where
+        K: std::borrow::Borrow<Q> + std::fmt::Debug,
+        Q: std::hash::Hash + Eq + std::fmt::Debug,
+    {
+        // Find the right segment
+        let branchseg_id = *self.branches.get(parent).with_context(|| {
+            format!(
+                "should had found the parent {:?} by key. in branches {:?}",
+                parent, self.branches
+            )
+        })?;
+
+        let _branchseg = &mut self.segments[branchseg_id];
+
+        // Create branch name for it
+        self.branches.insert(name, branchseg_id);
+        Ok(())
+    }
+
+    pub fn calculate(&mut self, retention_period: u64) -> anyhow::Result<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) -> anyhow::Result<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);
+        }
+        Ok(SegmentSize {
+            seg_id,
+            method: if seg.needed { WalNeeded } else { Wal },
+            this_size,
+            children,
+        })
+    }
+
+    fn size_from_snapshot_later(&self, seg_id: usize) -> anyhow::Result<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);
+            }
+            Ok(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)?);
+                }
+                let Some(this_size) = seg.end_size else { anyhow::bail!("no end_size at junction {seg_id}") };
+                Some(SegmentSize {
+                    seg_id,
+                    method: SnapshotAfter,
+                    this_size,
+                    children,
+                })
+            } else {
+                None
+            };
+
+            Ok(match (method1, method2) {
+                (None, None) => anyhow::bail!(
+                    "neither method was applicable: children_after={}, children_needed={}",
+                    seg.children_after.len(),
+                    children_needed
+                ),
+                (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 +1,269 @@
+//! 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() -> anyhow::Result<(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)?;
+
+    Ok((storage.into_segments(), size))
+}
+
+// Main branch only. Some updates on it.
+fn scenario_2() -> anyhow::Result<(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)?;
+
+    Ok((storage.into_segments(), size))
+}
+
+// Like 2, but more updates on main
+fn scenario_3() -> anyhow::Result<(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)?;
+
+    Ok((storage.into_segments(), size))
+}
+
+// Diverged branches
+fn scenario_4() -> anyhow::Result<(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)?;
+
+    Ok((storage.into_segments(), size))
+}
+
+fn scenario_5() -> anyhow::Result<(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)?;
+
+    Ok((storage.into_segments(), size))
+}
+
+fn scenario_6() -> anyhow::Result<(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)?;
+
+    Ok((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);
+        }
+    }
+    .unwrap();
+
+    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.unwrap_or(0);
+    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.unwrap_or(0) as i128 - seg.end_size.unwrap_or(0) 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() -> anyhow::Result<(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().unwrap();
+        assert_eq!(*expected, size.total_children(), "scenario on line {line}");
+    }
+}

libs/tenant_size_model/src/svg.rs

@@ -1,193 +0,0 @@
-use crate::{SegmentMethod, SegmentSizeResult, SizeResult, StorageModel};
-use std::fmt::Write;
-
-const SVG_WIDTH: f32 = 500.0;
-
-struct SvgDraw<'a> {
-    storage: &'a StorageModel,
-    branches: &'a [String],
-    seg_to_branch: &'a [usize],
-    sizes: &'a [SegmentSizeResult],
-
-    // layout
-    xscale: f32,
-    min_lsn: u64,
-    seg_coordinates: Vec<(f32, f32)>,
-}
-
-fn draw_legend(result: &mut String) -> anyhow::Result<()> {
-    writeln!(
-        result,
-        "<circle cx=\"10\" cy=\"10\" r=\"5\" stroke=\"red\"/>"
-    )?;
-    writeln!(result, "<text x=\"20\" y=\"15\">logical snapshot</text>")?;
-    writeln!(
-        result,
-        "<line x1=\"5\" y1=\"30\" x2=\"15\" y2=\"30\" stroke-width=\"6\" stroke=\"black\" />"
-    )?;
-    writeln!(
-        result,
-        "<text x=\"20\" y=\"35\">WAL within retention period</text>"
-    )?;
-    writeln!(
-        result,
-        "<line x1=\"5\" y1=\"50\" x2=\"15\" y2=\"50\" stroke-width=\"3\" stroke=\"black\" />"
-    )?;
-    writeln!(
-        result,
-        "<text x=\"20\" y=\"55\">WAL retained to avoid copy</text>"
-    )?;
-    writeln!(
-        result,
-        "<line x1=\"5\" y1=\"70\" x2=\"15\" y2=\"70\" stroke-width=\"1\" stroke=\"gray\" />"
-    )?;
-    writeln!(result, "<text x=\"20\" y=\"75\">WAL not retained</text>")?;
-    Ok(())
-}
-
-pub fn draw_svg(
-    storage: &StorageModel,
-    branches: &[String],
-    seg_to_branch: &[usize],
-    sizes: &SizeResult,
-) -> anyhow::Result<String> {
-    let mut draw = SvgDraw {
-        storage,
-        branches,
-        seg_to_branch,
-        sizes: &sizes.segments,
-
-        xscale: 0.0,
-        min_lsn: 0,
-        seg_coordinates: Vec::new(),
-    };
-
-    let mut result = String::new();
-
-    writeln!(result, "<svg xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\" height=\"300\" width=\"500\">")?;
-
-    draw.calculate_svg_layout();
-
-    // Draw the tree
-    for (seg_id, _seg) in storage.segments.iter().enumerate() {
-        draw.draw_seg_phase1(seg_id, &mut result)?;
-    }
-
-    // Draw snapshots
-    for (seg_id, _seg) in storage.segments.iter().enumerate() {
-        draw.draw_seg_phase2(seg_id, &mut result)?;
-    }
-
-    draw_legend(&mut result)?;
-
-    write!(result, "</svg>")?;
-
-    Ok(result)
-}
-
-impl<'a> SvgDraw<'a> {
-    fn calculate_svg_layout(&mut self) {
-        // Find x scale
-        let segments = &self.storage.segments;
-        let min_lsn = segments.iter().map(|s| s.lsn).fold(u64::MAX, std::cmp::min);
-        let max_lsn = segments.iter().map(|s| s.lsn).fold(0, std::cmp::max);
-
-        // Start with 1 pixel = 1 byte. Double the scale until it fits into the image
-        let mut xscale = 1.0;
-        while (max_lsn - min_lsn) as f32 / xscale > SVG_WIDTH {
-            xscale *= 2.0;
-        }
-
-        // Layout the timelines on Y dimension.
-        // TODO
-        let mut y = 100.0;
-        let mut branch_y_coordinates = Vec::new();
-        for _branch in self.branches {
-            branch_y_coordinates.push(y);
-            y += 40.0;
-        }
-
-        // Calculate coordinates for each point
-        let seg_coordinates = std::iter::zip(segments, self.seg_to_branch)
-            .map(|(seg, branch_id)| {
-                let x = (seg.lsn - min_lsn) as f32 / xscale;
-                let y = branch_y_coordinates[*branch_id];
-                (x, y)
-            })
-            .collect();
-
-        self.xscale = xscale;
-        self.min_lsn = min_lsn;
-        self.seg_coordinates = seg_coordinates;
-    }
-
-    /// Draws lines between points
-    fn draw_seg_phase1(&self, seg_id: usize, result: &mut String) -> anyhow::Result<()> {
-        let seg = &self.storage.segments[seg_id];
-
-        let wal_bytes = if let Some(parent_id) = seg.parent {
-            seg.lsn - self.storage.segments[parent_id].lsn
-        } else {
-            0
-        };
-
-        let style = match self.sizes[seg_id].method {
-            SegmentMethod::SnapshotHere => "stroke-width=\"1\" stroke=\"gray\"",
-            SegmentMethod::Wal if seg.needed && wal_bytes > 0 => {
-                "stroke-width=\"6\" stroke=\"black\""
-            }
-            SegmentMethod::Wal => "stroke-width=\"3\" stroke=\"black\"",
-            SegmentMethod::Skipped => "stroke-width=\"1\" stroke=\"gray\"",
-        };
-        if let Some(parent_id) = seg.parent {
-            let (x1, y1) = self.seg_coordinates[parent_id];
-            let (x2, y2) = self.seg_coordinates[seg_id];
-
-            writeln!(
-                result,
-                "<line x1=\"{x1}\" y1=\"{y1}\" x2=\"{x2}\" y2=\"{y2}\" {style}>",
-            )?;
-            writeln!(
-                result,
-                "  <title>{wal_bytes} bytes of WAL (seg {seg_id})</title>"
-            )?;
-            writeln!(result, "</line>")?;
-        } else {
-            // draw a little dash to mark the starting point of this branch
-            let (x, y) = self.seg_coordinates[seg_id];
-            let (x1, y1) = (x, y - 5.0);
-            let (x2, y2) = (x, y + 5.0);
-
-            writeln!(
-                result,
-                "<line x1=\"{x1}\" y1=\"{y1}\" x2=\"{x2}\" y2=\"{y2}\" {style}>",
-            )?;
-            writeln!(result, "  <title>(seg {seg_id})</title>")?;
-            writeln!(result, "</line>")?;
-        }
-
-        Ok(())
-    }
-
-    /// Draw circles where snapshots are taken
-    fn draw_seg_phase2(&self, seg_id: usize, result: &mut String) -> anyhow::Result<()> {
-        let seg = &self.storage.segments[seg_id];
-
-        // draw a snapshot point if it's needed
-        let (coord_x, coord_y) = self.seg_coordinates[seg_id];
-        if self.sizes[seg_id].method == SegmentMethod::SnapshotHere {
-            writeln!(
-                result,
-                "<circle cx=\"{coord_x}\" cy=\"{coord_y}\" r=\"5\" stroke=\"red\">",
-            )?;
-            writeln!(
-                result,
-                "  <title>logical size {}</title>",
-                seg.size.unwrap()
-            )?;
-            write!(result, "</circle>")?;
-        }
-
-        Ok(())
-    }
-}

libs/tenant_size_model/tests/tests.rs

@@ -1,313 +0,0 @@
-//! Tenant size model tests.
-
-use tenant_size_model::{Segment, SizeResult, StorageModel};
-
-use std::collections::HashMap;
-
-struct ScenarioBuilder {
-    segments: Vec<Segment>,
-
-    /// Mapping from the branch name to the index of a segment describing its latest state.
-    branches: HashMap<String, usize>,
-}
-
-impl ScenarioBuilder {
-    /// Creates a new storage with the given default branch name.
-    pub fn new(initial_branch: &str) -> ScenarioBuilder {
-        let init_segment = Segment {
-            parent: None,
-            lsn: 0,
-            size: Some(0),
-            needed: false, // determined later
-        };
-
-        ScenarioBuilder {
-            segments: vec![init_segment],
-            branches: HashMap::from([(initial_branch.into(), 0)]),
-        }
-    }
-
-    /// Advances the branch with the named operation, by the relative LSN and logical size bytes.
-    pub fn modify_branch(&mut self, branch: &str, lsn_bytes: u64, size_bytes: i64) {
-        let lastseg_id = *self.branches.get(branch).unwrap();
-        let newseg_id = self.segments.len();
-        let lastseg = &mut self.segments[lastseg_id];
-
-        let newseg = Segment {
-            parent: Some(lastseg_id),
-            lsn: lastseg.lsn + lsn_bytes,
-            size: Some((lastseg.size.unwrap() as i64 + size_bytes) as u64),
-            needed: false,
-        };
-
-        self.segments.push(newseg);
-        *self.branches.get_mut(branch).expect("read already") = newseg_id;
-    }
-
-    pub fn insert(&mut self, branch: &str, bytes: u64) {
-        self.modify_branch(branch, bytes, bytes as i64);
-    }
-
-    pub fn update(&mut self, branch: &str, bytes: u64) {
-        self.modify_branch(branch, bytes, 0i64);
-    }
-
-    pub fn _delete(&mut self, branch: &str, bytes: u64) {
-        self.modify_branch(branch, bytes, -(bytes as i64));
-    }
-
-    /// Panics if the parent branch cannot be found.
-    pub fn branch(&mut self, parent: &str, name: &str) {
-        // 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.to_string(), branchseg_id);
-    }
-
-    pub fn calculate(&mut self, retention_period: u64) -> (StorageModel, SizeResult) {
-        // 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.lsn.saturating_sub(retention_period);
-            let mut seg_id = last_seg_id;
-            loop {
-                let seg = &mut self.segments[seg_id];
-                if seg.lsn <= cutoff_lsn {
-                    break;
-                }
-                seg.needed = true;
-                if let Some(prev_seg_id) = seg.parent {
-                    seg_id = prev_seg_id;
-                } else {
-                    break;
-                }
-            }
-        }
-
-        // Perform the calculation
-        let storage_model = StorageModel {
-            segments: self.segments.clone(),
-        };
-        let size_result = storage_model.calculate();
-        (storage_model, size_result)
-    }
-}
-
-// Main branch only. Some updates on it.
-#[test]
-fn scenario_1() {
-    // Create main branch
-    let mut scenario = ScenarioBuilder::new("main");
-
-    // Bulk load 5 GB of data to it
-    scenario.insert("main", 5_000);
-
-    // Stream of updates
-    for _ in 0..5 {
-        scenario.update("main", 1_000);
-    }
-
-    // Calculate the synthetic size with retention horizon 1000
-    let (_model, result) = scenario.calculate(1000);
-
-    // The end of the branch is at LSN 10000. Need to retain
-    // a logical snapshot at LSN 9000, plus the WAL between 9000-10000.
-    // The logical snapshot has size 5000.
-    assert_eq!(result.total_size, 5000 + 1000);
-}
-
-// Main branch only. Some updates on it.
-#[test]
-fn scenario_2() {
-    // Create main branch
-    let mut scenario = ScenarioBuilder::new("main");
-
-    // Bulk load 5 GB of data to it
-    scenario.insert("main", 5_000);
-
-    // Stream of updates
-    for _ in 0..5 {
-        scenario.update("main", 1_000);
-    }
-
-    // Branch
-    scenario.branch("main", "child");
-    scenario.update("child", 1_000);
-
-    // More updates on parent
-    scenario.update("main", 1_000);
-
-    //
-    // The history looks like this now:
-    //
-    //         10000          11000
-    // *----*----*--------------*    main
-    //           |
-    //           |            11000
-    //           +--------------     child
-    //
-    //
-    // With retention horizon 1000, we need to retain logical snapshot
-    // at the branch point, size 5000, and the WAL from 10000-11000 on
-    // both branches.
-    let (_model, result) = scenario.calculate(1000);
-
-    assert_eq!(result.total_size, 5000 + 1000 + 1000);
-}
-
-// Like 2, but more updates on main
-#[test]
-fn scenario_3() {
-    // Create main branch
-    let mut scenario = ScenarioBuilder::new("main");
-
-    // Bulk load 5 GB of data to it
-    scenario.insert("main", 5_000);
-
-    // Stream of updates
-    for _ in 0..5 {
-        scenario.update("main", 1_000);
-    }
-
-    // Branch
-    scenario.branch("main", "child");
-    scenario.update("child", 1_000);
-
-    // More updates on parent
-    for _ in 0..5 {
-        scenario.update("main", 1_000);
-    }
-
-    //
-    // The history looks like this now:
-    //
-    //         10000                                 15000
-    // *----*----*------------------------------------*    main
-    //           |
-    //           |            11000
-    //           +--------------     child
-    //
-    //
-    // With retention horizon 1000, it's still cheapest to retain
-    // - snapshot at branch point (size 5000)
-    // - WAL on child between 10000-11000
-    // - WAL on main between 10000-15000
-    //
-    // This is in total 5000 + 1000 + 5000
-    //
-    let (_model, result) = scenario.calculate(1000);
-
-    assert_eq!(result.total_size, 5000 + 1000 + 5000);
-}
-
-// Diverged branches
-#[test]
-fn scenario_4() {
-    // Create main branch
-    let mut scenario = ScenarioBuilder::new("main");
-
-    // Bulk load 5 GB of data to it
-    scenario.insert("main", 5_000);
-
-    // Stream of updates
-    for _ in 0..5 {
-        scenario.update("main", 1_000);
-    }
-
-    // Branch
-    scenario.branch("main", "child");
-    scenario.update("child", 1_000);
-
-    // More updates on parent
-    for _ in 0..8 {
-        scenario.update("main", 1_000);
-    }
-
-    //
-    // The history looks like this now:
-    //
-    //         10000                                 18000
-    // *----*----*------------------------------------*    main
-    //           |
-    //           |            11000
-    //           +--------------     child
-    //
-    //
-    // With retention horizon 1000, it's now cheapest to retain
-    // separate snapshots on both branches:
-    // - snapshot on main branch at LSN 17000 (size 5000)
-    // - WAL on main between 17000-18000
-    // - snapshot on child branch at LSN 10000 (size 5000)
-    // - WAL on child between 10000-11000
-    //
-    // This is in total 5000 + 1000 + 5000 + 1000 = 12000
-    //
-    // (If we used the the method from the previous scenario, and
-    // kept only snapshot at the branch point, we'd need to keep
-    // all the WAL between 10000-18000 on the main branch, so
-    // the total size would be 5000 + 1000 + 8000 = 14000. The
-    // calculation always picks the cheapest alternative)
-
-    let (_model, result) = scenario.calculate(1000);
-
-    assert_eq!(result.total_size, 5000 + 1000 + 5000 + 1000);
-}
-
-#[test]
-fn scenario_5() {
-    let mut scenario = ScenarioBuilder::new("a");
-    scenario.insert("a", 5000);
-    scenario.branch("a", "b");
-    scenario.update("b", 4000);
-    scenario.update("a", 2000);
-    scenario.branch("a", "c");
-    scenario.insert("c", 4000);
-    scenario.insert("a", 2000);
-
-    let (_model, result) = scenario.calculate(1000);
-
-    assert_eq!(result.total_size, 17000);
-}
-
-#[test]
-fn scenario_6() {
-    let branches = [
-        "7ff1edab8182025f15ae33482edb590a",
-        "b1719e044db05401a05a2ed588a3ad3f",
-        "0xb68d6691c895ad0a70809470020929ef",
-    ];
-
-    // compared to other scenarios, this one uses bytes instead of kB
-
-    let mut scenario = ScenarioBuilder::new("");
-
-    scenario.branch("", branches[0]); // at 0
-    scenario.modify_branch(branches[0], 108951064, 43696128); // at 108951064
-    scenario.branch(branches[0], branches[1]); // at 108951064
-    scenario.modify_branch(branches[1], 15560408, -1851392); // at 124511472
-    scenario.modify_branch(branches[0], 174464360, -1531904); // at 283415424
-    scenario.branch(branches[0], branches[2]); // at 283415424
-    scenario.modify_branch(branches[2], 15906192, 8192); // at 299321616
-    scenario.modify_branch(branches[0], 18909976, 32768); // at 302325400
-
-    let (model, result) = scenario.calculate(100_000);
-
-    // FIXME: We previously calculated 333_792_000. But with this PR, we get
-    // a much lower number. At a quick look at the model output and the
-    // calculations here, the new result seems correct to me.
-    eprintln!(
-        " MODEL: {}",
-        serde_json::to_string(&model.segments).unwrap()
-    );
-    eprintln!(
-        "RESULT: {}",
-        serde_json::to_string(&result.segments).unwrap()
-    );
-
-    assert_eq!(result.total_size, 136_236_928);
-}

libs/tracing-utils/Cargo.toml

@@ -1,17 +0,0 @@
-[package]
-name = "tracing-utils"
-version = "0.1.0"
-edition.workspace = true
-license.workspace = true
-
-[dependencies]
-hyper.workspace = true
-opentelemetry = { workspace = true, features=["rt-tokio"] }
-opentelemetry-otlp = { workspace = true, default_features=false, features = ["http-proto", "trace", "http", "reqwest-client"] }
-opentelemetry-semantic-conventions.workspace = true
-reqwest = { workspace = true, default-features = false, features = ["rustls-tls"] }
-tokio = { workspace = true, features = ["rt", "rt-multi-thread"] }
-tracing.workspace = true
-tracing-opentelemetry.workspace = true
-tracing-subscriber.workspace = true
-workspace_hack = { version = "0.1", path = "../../workspace_hack" }

libs/tracing-utils/src/http.rs

@@ -1,96 +0,0 @@
-//! Tracing wrapper for Hyper HTTP server
-
-use hyper::HeaderMap;
-use hyper::{Body, Request, Response};
-use std::future::Future;
-use tracing::Instrument;
-use tracing_opentelemetry::OpenTelemetrySpanExt;
-
-/// Configuration option for what to use as the "otel.name" field in the traces.
-pub enum OtelName<'a> {
-    /// Use a constant string
-    Constant(&'a str),
-
-    /// Use the path from the request.
-    ///
-    /// That's very useful information, but is not appropriate if the
-    /// path contains parameters that differ on ever request, or worse,
-    /// sensitive information like usernames or email addresses.
-    ///
-    /// See <https://github.com/open-telemetry/opentelemetry-specification/blob/main/specification/trace/semantic_conventions/http.md#name>
-    UriPath,
-}
-
-/// Handle an incoming HTTP request using the given handler function,
-/// with OpenTelemetry tracing.
-///
-/// This runs 'handler' on the request in a new span, with fields filled in
-/// from the request. Notably, if the request contains tracing information,
-/// it is propagated to the span, so that this request is traced as part of
-/// the same trace.
-///
-/// XXX: Usually, this is handled by existing libraries, or built
-/// directly into HTTP servers. However, I couldn't find one for Hyper,
-/// so I had to write our own. OpenTelemetry website has a registry of
-/// instrumentation libraries at:
-/// https://opentelemetry.io/registry/?language=rust&component=instrumentation
-/// If a Hyper crate appears, consider switching to that.
-pub async fn tracing_handler<F, R>(
-    req: Request<Body>,
-    handler: F,
-    otel_name: OtelName<'_>,
-) -> Response<Body>
-where
-    F: Fn(Request<Body>) -> R,
-    R: Future<Output = Response<Body>>,
-{
-    // Create a tracing span, with context propagated from the incoming
-    // request if any.
-    //
-    // See list of standard fields defined for HTTP requests at
-    // https://github.com/open-telemetry/opentelemetry-specification/blob/main/specification/trace/semantic_conventions/http.md
-    // We only fill in a few of the most useful ones here.
-    let otel_name = match otel_name {
-        OtelName::Constant(s) => s,
-        OtelName::UriPath => req.uri().path(),
-    };
-
-    let span = tracing::info_span!(
-        "http request",
-        otel.name= %otel_name,
-        http.method = %req.method(),
-        http.status_code = tracing::field::Empty,
-    );
-    let parent_ctx = extract_remote_context(req.headers());
-    span.set_parent(parent_ctx);
-
-    // Handle the request within the span
-    let response = handler(req).instrument(span.clone()).await;
-
-    // Fill in the fields from the response code
-    let status = response.status();
-    span.record("http.status_code", status.as_str());
-    span.record(
-        "otel.status_code",
-        if status.is_success() { "OK" } else { "ERROR" },
-    );
-
-    response
-}
-
-// Extract remote tracing context from the HTTP headers
-fn extract_remote_context(headers: &HeaderMap) -> opentelemetry::Context {
-    struct HeaderExtractor<'a>(&'a HeaderMap);
-
-    impl<'a> opentelemetry::propagation::Extractor for HeaderExtractor<'a> {
-        fn get(&self, key: &str) -> Option<&str> {
-            self.0.get(key).and_then(|value| value.to_str().ok())
-        }
-
-        fn keys(&self) -> Vec<&str> {
-            self.0.keys().map(|value| value.as_str()).collect()
-        }
-    }
-    let extractor = HeaderExtractor(headers);
-    opentelemetry::global::get_text_map_propagator(|propagator| propagator.extract(&extractor))
-}

libs/tracing-utils/src/lib.rs

@@ -1,168 +0,0 @@
-//! Helper functions to set up OpenTelemetry tracing.
-//!
-//! This comes in two variants, depending on whether you have a Tokio runtime available.
-//! If you do, call `init_tracing()`. It sets up the trace processor and exporter to use
-//! the current tokio runtime. If you don't have a runtime available, or you don't want
-//! to share the runtime with the tracing tasks, call `init_tracing_without_runtime()`
-//! instead. It sets up a dedicated single-threaded Tokio runtime for the tracing tasks.
-//!
-//! Example:
-//!
-//! ```rust,no_run
-//! use tracing_subscriber::prelude::*;
-//! use tracing_opentelemetry::OpenTelemetryLayer;
-//!
-//! #[tokio::main]
-//! async fn main() {
-//!     // Set up logging to stderr
-//!     let env_filter = tracing_subscriber::EnvFilter::try_from_default_env()
-//!         .unwrap_or_else(|_| tracing_subscriber::EnvFilter::new("info"));
-//!     let fmt_layer = tracing_subscriber::fmt::layer()
-//!         .with_target(false)
-//!         .with_writer(std::io::stderr);
-//!
-//!     // Initialize OpenTelemetry. Exports tracing spans as OpenTelemetry traces
-//!     let otlp_layer = tracing_utils::init_tracing("my_application").await.map(OpenTelemetryLayer::new);
-//!
-//!     // Put it all together
-//!     tracing_subscriber::registry()
-//!         .with(env_filter)
-//!         .with(otlp_layer)
-//!         .with(fmt_layer)
-//!         .init();
-//! }
-//! ```
-
-use opentelemetry::sdk::Resource;
-use opentelemetry::KeyValue;
-use opentelemetry_otlp::WithExportConfig;
-use opentelemetry_otlp::{OTEL_EXPORTER_OTLP_ENDPOINT, OTEL_EXPORTER_OTLP_TRACES_ENDPOINT};
-
-pub use tracing_opentelemetry::OpenTelemetryLayer;
-
-pub mod http;
-
-/// Set up OpenTelemetry exporter, using configuration from environment variables.
-///
-/// `service_name` is set as the OpenTelemetry 'service.name' resource (see
-/// <https://github.com/open-telemetry/opentelemetry-specification/blob/main/specification/resource/semantic_conventions/README.md#service>)
-///
-/// We try to follow the conventions for the environment variables specified in
-/// <https://opentelemetry.io/docs/reference/specification/sdk-environment-variables/>
-///
-/// However, we only support a subset of those options:
-///
-/// - OTEL_SDK_DISABLED is supported. The default is "false", meaning tracing
-///   is enabled by default. Set it to "true" to disable.
-///
-/// - We use the OTLP exporter, with HTTP protocol. Most of the OTEL_EXPORTER_OTLP_*
-///   settings specified in
-///   <https://opentelemetry.io/docs/reference/specification/protocol/exporter/>
-///   are supported, as they are handled by the `opentelemetry-otlp` crate.
-///   Settings related to other exporters have no effect.
-///
-/// - Some other settings are supported by the `opentelemetry` crate.
-///
-/// If you need some other setting, please test if it works first. And perhaps
-/// add a comment in the list above to save the effort of testing for the next
-/// person.
-///
-/// This doesn't block, but is marked as 'async' to hint that this must be called in
-/// asynchronous execution context.
-pub async fn init_tracing(service_name: &str) -> Option<opentelemetry::sdk::trace::Tracer> {
-    if std::env::var("OTEL_SDK_DISABLED") == Ok("true".to_string()) {
-        return None;
-    };
-    Some(init_tracing_internal(service_name.to_string()))
-}
-
-/// Like `init_tracing`, but creates a separate tokio Runtime for the tracing
-/// tasks.
-pub fn init_tracing_without_runtime(
-    service_name: &str,
-) -> Option<opentelemetry::sdk::trace::Tracer> {
-    if std::env::var("OTEL_SDK_DISABLED") == Ok("true".to_string()) {
-        return None;
-    };
-
-    // The opentelemetry batch processor and the OTLP exporter needs a Tokio
-    // runtime. Create a dedicated runtime for them. One thread should be
-    // enough.
-    //
-    // (Alternatively, instead of batching, we could use the "simple
-    // processor", which doesn't need Tokio, and use "reqwest-blocking"
-    // feature for the OTLP exporter, which also doesn't need Tokio.  However,
-    // batching is considered best practice, and also I have the feeling that
-    // the non-Tokio codepaths in the opentelemetry crate are less used and
-    // might be more buggy, so better to stay on the well-beaten path.)
-    //
-    // We leak the runtime so that it keeps running after we exit the
-    // function.
-    let runtime = Box::leak(Box::new(
-        tokio::runtime::Builder::new_multi_thread()
-            .enable_all()
-            .thread_name("otlp runtime thread")
-            .worker_threads(1)
-            .build()
-            .unwrap(),
-    ));
-    let _guard = runtime.enter();
-
-    Some(init_tracing_internal(service_name.to_string()))
-}
-
-fn init_tracing_internal(service_name: String) -> opentelemetry::sdk::trace::Tracer {
-    // Set up exporter from the OTEL_EXPORTER_* environment variables
-    let mut exporter = opentelemetry_otlp::new_exporter().http().with_env();
-
-    // XXX opentelemetry-otlp v0.18.0 has a bug in how it uses the
-    // OTEL_EXPORTER_OTLP_ENDPOINT env variable. According to the
-    // OpenTelemetry spec at
-    // <https://github.com/open-telemetry/opentelemetry-specification/blob/main/specification/protocol/exporter.md#endpoint-urls-for-otlphttp>,
-    // the full exporter URL is formed by appending "/v1/traces" to the value
-    // of OTEL_EXPORTER_OTLP_ENDPOINT. However, opentelemetry-otlp only does
-    // that with the grpc-tonic exporter. Other exporters, like the HTTP
-    // exporter, use the URL from OTEL_EXPORTER_OTLP_ENDPOINT as is, without
-    // appending "/v1/traces".
-    //
-    // See https://github.com/open-telemetry/opentelemetry-rust/pull/950
-    //
-    // Work around that by checking OTEL_EXPORTER_OTLP_ENDPOINT, and setting
-    // the endpoint url with the "/v1/traces" path ourselves. If the bug is
-    // fixed in a later version, we can remove this code. But if we don't
-    // remember to remove this, it won't do any harm either, as the crate will
-    // just ignore the OTEL_EXPORTER_OTLP_ENDPOINT setting when the endpoint
-    // is set directly with `with_endpoint`.
-    if std::env::var(OTEL_EXPORTER_OTLP_TRACES_ENDPOINT).is_err() {
-        if let Ok(mut endpoint) = std::env::var(OTEL_EXPORTER_OTLP_ENDPOINT) {
-            if !endpoint.ends_with('/') {
-                endpoint.push('/');
-            }
-            endpoint.push_str("v1/traces");
-            exporter = exporter.with_endpoint(endpoint);
-        }
-    }
-
-    // Propagate trace information in the standard W3C TraceContext format.
-    opentelemetry::global::set_text_map_propagator(
-        opentelemetry::sdk::propagation::TraceContextPropagator::new(),
-    );
-
-    opentelemetry_otlp::new_pipeline()
-        .tracing()
-        .with_exporter(exporter)
-        .with_trace_config(
-            opentelemetry::sdk::trace::config().with_resource(Resource::new(vec![KeyValue::new(
-                opentelemetry_semantic_conventions::resource::SERVICE_NAME,
-                service_name,
-            )])),
-        )
-        .install_batch(opentelemetry::runtime::Tokio)
-        .expect("could not initialize opentelemetry exporter")
-}
-
-// Shutdown trace pipeline gracefully, so that it has a chance to send any
-// pending traces before we exit.
-pub fn shutdown_tracing() {
-    opentelemetry::global::shutdown_tracer_provider();
-}

libs/utils/Cargo.toml

@@ -11,38 +11,40 @@ async-trait.workspace = true
 anyhow.workspace = true
 bincode.workspace = true
 bytes.workspace = true
-heapless.workspace = true
-hex = { workspace = true, features = ["serde"] }
 hyper = { workspace = true, features = ["full"] }
-futures = { workspace = true}
-jsonwebtoken.workspace = true
-nix.workspace = true
-once_cell.workspace = true
-pin-project-lite.workspace = true
 routerify.workspace = true
 serde.workspace = true
 serde_json.workspace = true
-signal-hook.workspace = true
 thiserror.workspace = true
 tokio.workspace = true
+tokio-rustls.workspace = true
 tracing.workspace = true
 tracing-subscriber = { workspace = true, features = ["json"] }
+nix.workspace = true
+signal-hook.workspace = true
 rand.workspace = true
+jsonwebtoken.workspace = true
+hex = { workspace = true, features = ["serde"] }
+rustls.workspace = true
+rustls-split.workspace = true
+git-version.workspace = true
 serde_with.workspace = true
+once_cell.workspace = true
 strum.workspace = true
 strum_macros.workspace = true
-url.workspace = true
-uuid = { version = "1.2", features = ["v4", "serde"] }
 
 metrics.workspace = true
+pq_proto.workspace = true
+
 workspace_hack.workspace = true
 
 [dev-dependencies]
 byteorder.workspace = true
 bytes.workspace = true
-criterion.workspace = true
 hex-literal.workspace = true
 tempfile.workspace = true
+criterion.workspace = true
+rustls-pemfile.workspace = true
 
 [[bench]]
 name = "benchmarks"

libs/utils/src/auth.rs

@@ -1,4 +1,7 @@
 // For details about authentication see docs/authentication.md
+//
+// TODO: use ed25519 keys
+// Relevant issue: https://github.com/Keats/jsonwebtoken/issues/162
 
 use serde;
 use std::fs;
@@ -13,10 +16,9 @@ use serde_with::{serde_as, DisplayFromStr};
 
 use crate::id::TenantId;
 
-/// Algorithm to use. We require EdDSA.
-const STORAGE_TOKEN_ALGORITHM: Algorithm = Algorithm::EdDSA;
+const JWT_ALGORITHM: Algorithm = Algorithm::RS256;
 
-#[derive(Debug, Serialize, Deserialize, Clone, PartialEq)]
+#[derive(Debug, Serialize, Deserialize, Clone)]
 #[serde(rename_all = "lowercase")]
 pub enum Scope {
     // Provides access to all data for a specific tenant (specified in `struct Claims` below)
@@ -31,9 +33,8 @@ pub enum Scope {
     SafekeeperData,
 }
 
-/// JWT payload. See docs/authentication.md for the format
 #[serde_as]
-#[derive(Debug, Serialize, Deserialize, Clone, PartialEq)]
+#[derive(Debug, Serialize, Deserialize, Clone)]
 pub struct Claims {
     #[serde(default)]
     #[serde_as(as = "Option<DisplayFromStr>")]
@@ -54,8 +55,7 @@ pub struct JwtAuth {
 
 impl JwtAuth {
     pub fn new(decoding_key: DecodingKey) -> Self {
-        let mut validation = Validation::default();
-        validation.algorithms = vec![STORAGE_TOKEN_ALGORITHM];
+        let mut validation = Validation::new(JWT_ALGORITHM);
         // The default 'required_spec_claims' is 'exp'. But we don't want to require
         // expiration.
         validation.required_spec_claims = [].into();
@@ -67,7 +67,7 @@ impl JwtAuth {
 
     pub fn from_key_path(key_path: &Path) -> Result<Self> {
         let public_key = fs::read(key_path)?;
-        Ok(Self::new(DecodingKey::from_ed_pem(&public_key)?))
+        Ok(Self::new(DecodingKey::from_rsa_pem(&public_key)?))
     }
 
     pub fn decode(&self, token: &str) -> Result<TokenData<Claims>> {
@@ -85,75 +85,6 @@ impl std::fmt::Debug for JwtAuth {
 
 // this function is used only for testing purposes in CLI e g generate tokens during init
 pub fn encode_from_key_file(claims: &Claims, key_data: &[u8]) -> Result<String> {
-    let key = EncodingKey::from_ed_pem(key_data)?;
-    Ok(encode(&Header::new(STORAGE_TOKEN_ALGORITHM), claims, &key)?)
-}
-
-#[cfg(test)]
-mod tests {
-    use super::*;
-    use std::str::FromStr;
-
-    // Generated with:
-    //
-    // openssl genpkey -algorithm ed25519 -out ed25519-priv.pem
-    // openssl pkey -in ed25519-priv.pem -pubout -out ed25519-pub.pem
-    const TEST_PUB_KEY_ED25519: &[u8] = br#"
------BEGIN PUBLIC KEY-----
-MCowBQYDK2VwAyEARYwaNBayR+eGI0iXB4s3QxE3Nl2g1iWbr6KtLWeVD/w=
------END PUBLIC KEY-----
-"#;
-
-    const TEST_PRIV_KEY_ED25519: &[u8] = br#"
------BEGIN PRIVATE KEY-----
-MC4CAQAwBQYDK2VwBCIEID/Drmc1AA6U/znNRWpF3zEGegOATQxfkdWxitcOMsIH
------END PRIVATE KEY-----
-"#;
-
-    #[test]
-    fn test_decode() -> Result<(), anyhow::Error> {
-        let expected_claims = Claims {
-            tenant_id: Some(TenantId::from_str("3d1f7595b468230304e0b73cecbcb081")?),
-            scope: Scope::Tenant,
-        };
-
-        // A test token containing the following payload, signed using TEST_PRIV_KEY_ED25519:
-        //
-        // ```
-        // {
-        //   "scope": "tenant",
-        //   "tenant_id": "3d1f7595b468230304e0b73cecbcb081",
-        //   "iss": "neon.controlplane",
-        //   "exp": 1709200879,
-        //   "iat": 1678442479
-        // }
-        // ```
-        //
-        let encoded_eddsa = "eyJhbGciOiJFZERTQSIsInR5cCI6IkpXVCJ9.eyJzY29wZSI6InRlbmFudCIsInRlbmFudF9pZCI6IjNkMWY3NTk1YjQ2ODIzMDMwNGUwYjczY2VjYmNiMDgxIiwiaXNzIjoibmVvbi5jb250cm9scGxhbmUiLCJleHAiOjE3MDkyMDA4NzksImlhdCI6MTY3ODQ0MjQ3OX0.U3eA8j-uU-JnhzeO3EDHRuXLwkAUFCPxtGHEgw6p7Ccc3YRbFs2tmCdbD9PZEXP-XsxSeBQi1FY0YPcT3NXADw";
-
-        // Check it can be validated with the public key
-        let auth = JwtAuth::new(DecodingKey::from_ed_pem(TEST_PUB_KEY_ED25519)?);
-        let claims_from_token = auth.decode(encoded_eddsa)?.claims;
-        assert_eq!(claims_from_token, expected_claims);
-
-        Ok(())
-    }
-
-    #[test]
-    fn test_encode() -> Result<(), anyhow::Error> {
-        let claims = Claims {
-            tenant_id: Some(TenantId::from_str("3d1f7595b468230304e0b73cecbcb081")?),
-            scope: Scope::Tenant,
-        };
-
-        let encoded = encode_from_key_file(&claims, TEST_PRIV_KEY_ED25519)?;
-
-        // decode it back
-        let auth = JwtAuth::new(DecodingKey::from_ed_pem(TEST_PUB_KEY_ED25519)?);
-        let decoded = auth.decode(&encoded)?;
-
-        assert_eq!(decoded.claims, claims);
-
-        Ok(())
-    }
+    let key = EncodingKey::from_rsa_pem(key_data)?;
+    Ok(encode(&Header::new(JWT_ALGORITHM), claims, &key)?)
 }

libs/utils/src/fs_ext.rs

@@ -11,7 +11,7 @@ where
     P: AsRef<Path>,
 {
     fn is_empty_dir(&self) -> io::Result<bool> {
-        Ok(fs::read_dir(self)?.next().is_none())
+        Ok(fs::read_dir(self)?.into_iter().next().is_none())
     }
 }
 

libs/utils/src/history_buffer.rs

@@ -1,161 +0,0 @@
-//! A heapless buffer for events of sorts.
-
-use std::ops;
-
-use heapless::HistoryBuffer;
-
-#[derive(Debug, Clone)]
-pub struct HistoryBufferWithDropCounter<T, const L: usize> {
-    buffer: HistoryBuffer<T, L>,
-    drop_count: u64,
-}
-
-impl<T, const L: usize> HistoryBufferWithDropCounter<T, L> {
-    pub fn write(&mut self, data: T) {
-        let len_before = self.buffer.len();
-        self.buffer.write(data);
-        let len_after = self.buffer.len();
-        self.drop_count += u64::from(len_before == len_after);
-    }
-    pub fn drop_count(&self) -> u64 {
-        self.drop_count
-    }
-    pub fn map<U, F: Fn(&T) -> U>(&self, f: F) -> HistoryBufferWithDropCounter<U, L> {
-        let mut buffer = HistoryBuffer::new();
-        buffer.extend(self.buffer.oldest_ordered().map(f));
-        HistoryBufferWithDropCounter::<U, L> {
-            buffer,
-            drop_count: self.drop_count,
-        }
-    }
-}
-
-impl<T, const L: usize> Default for HistoryBufferWithDropCounter<T, L> {
-    fn default() -> Self {
-        Self {
-            buffer: HistoryBuffer::default(),
-            drop_count: 0,
-        }
-    }
-}
-
-impl<T, const L: usize> ops::Deref for HistoryBufferWithDropCounter<T, L> {
-    type Target = HistoryBuffer<T, L>;
-
-    fn deref(&self) -> &Self::Target {
-        &self.buffer
-    }
-}
-
-#[derive(serde::Serialize)]
-struct SerdeRepr<T> {
-    buffer: Vec<T>,
-    drop_count: u64,
-}
-
-impl<'a, T, const L: usize> From<&'a HistoryBufferWithDropCounter<T, L>> for SerdeRepr<T>
-where
-    T: Clone + serde::Serialize,
-{
-    fn from(value: &'a HistoryBufferWithDropCounter<T, L>) -> Self {
-        let HistoryBufferWithDropCounter { buffer, drop_count } = value;
-        SerdeRepr {
-            buffer: buffer.iter().cloned().collect(),
-            drop_count: *drop_count,
-        }
-    }
-}
-
-impl<T, const L: usize> serde::Serialize for HistoryBufferWithDropCounter<T, L>
-where
-    T: Clone + serde::Serialize,
-{
-    fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
-    where
-        S: serde::Serializer,
-    {
-        SerdeRepr::from(self).serialize(serializer)
-    }
-}
-
-#[cfg(test)]
-mod test {
-    use super::HistoryBufferWithDropCounter;
-
-    #[test]
-    fn test_basics() {
-        let mut b = HistoryBufferWithDropCounter::<_, 2>::default();
-        b.write(1);
-        b.write(2);
-        b.write(3);
-        assert!(b.iter().any(|e| *e == 2));
-        assert!(b.iter().any(|e| *e == 3));
-        assert!(!b.iter().any(|e| *e == 1));
-    }
-
-    #[test]
-    fn test_drop_count_works() {
-        let mut b = HistoryBufferWithDropCounter::<_, 2>::default();
-        b.write(1);
-        assert_eq!(b.drop_count(), 0);
-        b.write(2);
-        assert_eq!(b.drop_count(), 0);
-        b.write(3);
-        assert_eq!(b.drop_count(), 1);
-        b.write(4);
-        assert_eq!(b.drop_count(), 2);
-    }
-
-    #[test]
-    fn test_clone_works() {
-        let mut b = HistoryBufferWithDropCounter::<_, 2>::default();
-        b.write(1);
-        b.write(2);
-        b.write(3);
-        assert_eq!(b.drop_count(), 1);
-        let mut c = b.clone();
-        assert_eq!(c.drop_count(), 1);
-        assert!(c.iter().any(|e| *e == 2));
-        assert!(c.iter().any(|e| *e == 3));
-        assert!(!c.iter().any(|e| *e == 1));
-
-        c.write(4);
-        assert!(c.iter().any(|e| *e == 4));
-        assert!(!b.iter().any(|e| *e == 4));
-    }
-
-    #[test]
-    fn test_map() {
-        let mut b = HistoryBufferWithDropCounter::<_, 2>::default();
-
-        b.write(1);
-        assert_eq!(b.drop_count(), 0);
-        {
-            let c = b.map(|i| i + 10);
-            assert_eq!(c.oldest_ordered().cloned().collect::<Vec<_>>(), vec![11]);
-            assert_eq!(c.drop_count(), 0);
-        }
-
-        b.write(2);
-        assert_eq!(b.drop_count(), 0);
-        {
-            let c = b.map(|i| i + 10);
-            assert_eq!(
-                c.oldest_ordered().cloned().collect::<Vec<_>>(),
-                vec![11, 12]
-            );
-            assert_eq!(c.drop_count(), 0);
-        }
-
-        b.write(3);
-        assert_eq!(b.drop_count(), 1);
-        {
-            let c = b.map(|i| i + 10);
-            assert_eq!(
-                c.oldest_ordered().cloned().collect::<Vec<_>>(),
-                vec![12, 13]
-            );
-            assert_eq!(c.drop_count(), 1);
-        }
-    }
-}

libs/utils/src/http/endpoint.rs

@@ -1,20 +1,18 @@
 use crate::auth::{Claims, JwtAuth};
 use crate::http::error;
-use anyhow::{anyhow, Context};
-use hyper::header::{HeaderName, AUTHORIZATION};
-use hyper::http::HeaderValue;
-use hyper::Method;
+use anyhow::anyhow;
+use hyper::header::AUTHORIZATION;
 use hyper::{header::CONTENT_TYPE, Body, Request, Response, Server};
 use metrics::{register_int_counter, Encoder, IntCounter, TextEncoder};
 use once_cell::sync::Lazy;
 use routerify::ext::RequestExt;
-use routerify::{Middleware, RequestInfo, Router, RouterBuilder, RouterService};
+use routerify::RequestInfo;
+use routerify::{Middleware, Router, RouterBuilder, RouterService};
 use tokio::task::JoinError;
-use tracing::{self, debug, info, info_span, warn, Instrument};
+use tracing::info;
 
 use std::future::Future;
 use std::net::TcpListener;
-use std::str::FromStr;
 
 use super::error::ApiError;
 
@@ -26,83 +24,9 @@ static SERVE_METRICS_COUNT: Lazy<IntCounter> = Lazy::new(|| {
     .expect("failed to define a metric")
 });
 
-static X_REQUEST_ID_HEADER_STR: &str = "x-request-id";
-
-static X_REQUEST_ID_HEADER: HeaderName = HeaderName::from_static(X_REQUEST_ID_HEADER_STR);
-#[derive(Debug, Default, Clone)]
-struct RequestId(String);
-
-/// Adds a tracing info_span! instrumentation around the handler events,
-/// logs the request start and end events for non-GET requests and non-200 responses.
-///
-/// Use this to distinguish between logs of different HTTP requests: every request handler wrapped
-/// in this type will get request info logged in the wrapping span, including the unique request ID.
-///
-/// There could be other ways to implement similar functionality:
-///
-/// * procmacros placed on top of all handler methods
-/// With all the drawbacks of procmacros, brings no difference implementation-wise,
-/// and little code reduction compared to the existing approach.
-///
-/// * Another `TraitExt` with e.g. the `get_with_span`, `post_with_span` methods to do similar logic,
-/// implemented for [`RouterBuilder`].
-/// Could be simpler, but we don't want to depend on [`routerify`] more, targeting to use other library later.
-///
-/// * In theory, a span guard could've been created in a pre-request middleware and placed into a global collection, to be dropped
-/// later, in a post-response middleware.
-/// Due to suspendable nature of the futures, would give contradictive results which is exactly the opposite of what `tracing-futures`
-/// tries to achive with its `.instrument` used in the current approach.
-///
-/// If needed, a declarative macro to substitute the |r| ... closure boilerplate could be introduced.
-pub struct RequestSpan<E, R, H>(pub H)
-where
-    E: Into<Box<dyn std::error::Error + Send + Sync>> + 'static,
-    R: Future<Output = Result<Response<Body>, E>> + Send + 'static,
-    H: Fn(Request<Body>) -> R + Send + Sync + 'static;
-
-impl<E, R, H> RequestSpan<E, R, H>
-where
-    E: Into<Box<dyn std::error::Error + Send + Sync>> + 'static,
-    R: Future<Output = Result<Response<Body>, E>> + Send + 'static,
-    H: Fn(Request<Body>) -> R + Send + Sync + 'static,
-{
-    /// Creates a tracing span around inner request handler and executes the request handler in the contex of that span.
-    /// Use as `|r| RequestSpan(my_handler).handle(r)` instead of `my_handler` as the request handler to get the span enabled.
-    pub async fn handle(self, request: Request<Body>) -> Result<Response<Body>, E> {
-        let request_id = request.context::<RequestId>().unwrap_or_default().0;
-        let method = request.method();
-        let path = request.uri().path();
-        let request_span = info_span!("request", %method, %path, %request_id);
-
-        let log_quietly = method == Method::GET;
-        async move {
-            if log_quietly {
-                debug!("Handling request");
-            } else {
-                info!("Handling request");
-            }
-
-            // Note that we reuse `error::handler` here and not returning and error at all,
-            // yet cannot use `!` directly in the method signature due to `routerify::RouterBuilder` limitation.
-            // Usage of the error handler also means that we expect only the `ApiError` errors to be raised in this call.
-            //
-            // Panics are not handled separately, there's a `tracing_panic_hook` from another module to do that globally.
-            match (self.0)(request).await {
-                Ok(response) => {
-                    let response_status = response.status();
-                    if log_quietly && response_status.is_success() {
-                        debug!("Request handled, status: {response_status}");
-                    } else {
-                        info!("Request handled, status: {response_status}");
-                    }
-                    Ok(response)
-                }
-                Err(e) => Ok(error::handler(e.into()).await),
-            }
-        }
-        .instrument(request_span)
-        .await
-    }
+async fn logger(res: Response<Body>, info: RequestInfo) -> Result<Response<Body>, ApiError> {
+    info!("{} {} {}", info.method(), info.uri().path(), res.status(),);
+    Ok(res)
 }
 
 async fn prometheus_metrics_handler(_req: Request<Body>) -> Result<Response<Body>, ApiError> {
@@ -129,48 +53,10 @@ async fn prometheus_metrics_handler(_req: Request<Body>) -> Result<Response<Body
     Ok(response)
 }
 
-pub fn add_request_id_middleware<B: hyper::body::HttpBody + Send + Sync + 'static>(
-) -> Middleware<B, ApiError> {
-    Middleware::pre(move |req| async move {
-        let request_id = match req.headers().get(&X_REQUEST_ID_HEADER) {
-            Some(request_id) => request_id
-                .to_str()
-                .expect("extract request id value")
-                .to_owned(),
-            None => {
-                let request_id = uuid::Uuid::new_v4();
-                request_id.to_string()
-            }
-        };
-        req.set_context(RequestId(request_id));
-
-        Ok(req)
-    })
-}
-
-async fn add_request_id_header_to_response(
-    mut res: Response<Body>,
-    req_info: RequestInfo,
-) -> Result<Response<Body>, ApiError> {
-    if let Some(request_id) = req_info.context::<RequestId>() {
-        if let Ok(request_header_value) = HeaderValue::from_str(&request_id.0) {
-            res.headers_mut()
-                .insert(&X_REQUEST_ID_HEADER, request_header_value);
-        };
-    };
-
-    Ok(res)
-}
-
 pub fn make_router() -> RouterBuilder<hyper::Body, ApiError> {
     Router::builder()
-        .middleware(add_request_id_middleware())
-        .middleware(Middleware::post_with_info(
-            add_request_id_header_to_response,
-        ))
-        .get("/metrics", |r| {
-            RequestSpan(prometheus_metrics_handler).handle(r)
-        })
+        .middleware(Middleware::post_with_info(logger))
+        .get("/metrics", prometheus_metrics_handler)
         .err_handler(error::handler)
 }
 
@@ -180,43 +66,40 @@ pub fn attach_openapi_ui(
     spec_mount_path: &'static str,
     ui_mount_path: &'static str,
 ) -> RouterBuilder<hyper::Body, ApiError> {
-    router_builder
-        .get(spec_mount_path, move |r| {
-            RequestSpan(move |_| async move { Ok(Response::builder().body(Body::from(spec)).unwrap()) })
-                .handle(r)
-        })
-        .get(ui_mount_path, move |r| RequestSpan( move |_| async move {
-            Ok(Response::builder().body(Body::from(format!(r#"
-                <!DOCTYPE html>
-                <html lang="en">
-                <head>
-                <title>rweb</title>
-                <link href="https://cdn.jsdelivr.net/npm/swagger-ui-dist@3/swagger-ui.css" rel="stylesheet">
-                </head>
-                <body>
-                    <div id="swagger-ui"></div>
-                    <script src="https://cdn.jsdelivr.net/npm/swagger-ui-dist@3/swagger-ui-bundle.js" charset="UTF-8"> </script>
-                    <script>
-                        window.onload = function() {{
-                        const ui = SwaggerUIBundle({{
-                            "dom_id": "\#swagger-ui",
-                            presets: [
-                            SwaggerUIBundle.presets.apis,
-                            SwaggerUIBundle.SwaggerUIStandalonePreset
-                            ],
-                            layout: "BaseLayout",
-                            deepLinking: true,
-                            showExtensions: true,
-                            showCommonExtensions: true,
-                            url: "{}",
-                        }})
-                        window.ui = ui;
-                    }};
-                </script>
-                </body>
-                </html>
-            "#, spec_mount_path))).unwrap())
-        }).handle(r))
+    router_builder.get(spec_mount_path, move |_| async move {
+        Ok(Response::builder().body(Body::from(spec)).unwrap())
+    }).get(ui_mount_path, move |_| async move {
+        Ok(Response::builder().body(Body::from(format!(r#"
+            <!DOCTYPE html>
+            <html lang="en">
+            <head>
+            <title>rweb</title>
+            <link href="https://cdn.jsdelivr.net/npm/swagger-ui-dist@3/swagger-ui.css" rel="stylesheet">
+            </head>
+            <body>
+                <div id="swagger-ui"></div>
+                <script src="https://cdn.jsdelivr.net/npm/swagger-ui-dist@3/swagger-ui-bundle.js" charset="UTF-8"> </script>
+                <script>
+                    window.onload = function() {{
+                    const ui = SwaggerUIBundle({{
+                        "dom_id": "\#swagger-ui",
+                        presets: [
+                        SwaggerUIBundle.presets.apis,
+                        SwaggerUIBundle.SwaggerUIStandalonePreset
+                        ],
+                        layout: "BaseLayout",
+                        deepLinking: true,
+                        showExtensions: true,
+                        showCommonExtensions: true,
+                        url: "{}",
+                    }})
+                    window.ui = ui;
+                }};
+            </script>
+            </body>
+            </html>
+        "#, spec_mount_path))).unwrap())
+    })
 }
 
 fn parse_token(header_value: &str) -> Result<&str, ApiError> {
@@ -260,38 +143,6 @@ pub fn auth_middleware<B: hyper::body::HttpBody + Send + Sync + 'static>(
     })
 }
 
-pub fn add_response_header_middleware<B>(
-    header: &str,
-    value: &str,
-) -> anyhow::Result<Middleware<B, ApiError>>
-where
-    B: hyper::body::HttpBody + Send + Sync + 'static,
-{
-    let name =
-        HeaderName::from_str(header).with_context(|| format!("invalid header name: {header}"))?;
-    let value =
-        HeaderValue::from_str(value).with_context(|| format!("invalid header value: {value}"))?;
-    Ok(Middleware::post_with_info(
-        move |mut response, request_info| {
-            let name = name.clone();
-            let value = value.clone();
-            async move {
-                let headers = response.headers_mut();
-                if headers.contains_key(&name) {
-                    warn!(
-                        "{} response already contains header {:?}",
-                        request_info.uri(),
-                        &name,
-                    );
-                } else {
-                    headers.insert(name, value);
-                }
-                Ok(response)
-            }
-        },
-    ))
-}
-
 pub fn check_permission_with(
     req: &Request<Body>,
     check_permission: impl Fn(&Claims) -> Result<(), anyhow::Error>,
@@ -338,48 +189,3 @@ where
 
     Ok(())
 }
-#[cfg(test)]
-mod tests {
-    use super::*;
-    use futures::future::poll_fn;
-    use hyper::service::Service;
-    use routerify::RequestServiceBuilder;
-    use std::net::{IpAddr, SocketAddr};
-
-    #[tokio::test]
-    async fn test_request_id_returned() {
-        let builder = RequestServiceBuilder::new(make_router().build().unwrap()).unwrap();
-        let remote_addr = SocketAddr::new(IpAddr::from_str("127.0.0.1").unwrap(), 80);
-        let mut service = builder.build(remote_addr);
-        if let Err(e) = poll_fn(|ctx| service.poll_ready(ctx)).await {
-            panic!("request service is not ready: {:?}", e);
-        }
-
-        let mut req: Request<Body> = Request::default();
-        req.headers_mut()
-            .append(&X_REQUEST_ID_HEADER, HeaderValue::from_str("42").unwrap());
-
-        let resp: Response<hyper::body::Body> = service.call(req).await.unwrap();
-
-        let header_val = resp.headers().get(&X_REQUEST_ID_HEADER).unwrap();
-
-        assert!(header_val == "42", "response header mismatch");
-    }
-
-    #[tokio::test]
-    async fn test_request_id_empty() {
-        let builder = RequestServiceBuilder::new(make_router().build().unwrap()).unwrap();
-        let remote_addr = SocketAddr::new(IpAddr::from_str("127.0.0.1").unwrap(), 80);
-        let mut service = builder.build(remote_addr);
-        if let Err(e) = poll_fn(|ctx| service.poll_ready(ctx)).await {
-            panic!("request service is not ready: {:?}", e);
-        }
-
-        let req: Request<Body> = Request::default();
-        let resp: Response<hyper::body::Body> = service.call(req).await.unwrap();
-
-        let header_val = resp.headers().get(&X_REQUEST_ID_HEADER);
-
-        assert_ne!(header_val, None, "response header should NOT be empty");
-    }
-}

libs/utils/src/http/error.rs

@@ -1,7 +1,6 @@
 use hyper::{header, Body, Response, StatusCode};
 use serde::{Deserialize, Serialize};
 use thiserror::Error;
-use tracing::error;
 
 #[derive(Debug, Error)]
 pub enum ApiError {
@@ -77,16 +76,8 @@ impl HttpErrorBody {
 }
 
 pub async fn handler(err: routerify::RouteError) -> Response<Body> {
-    let api_error = err
-        .downcast::<ApiError>()
-        .expect("handler should always return api error");
-
-    // Print a stack trace for Internal Server errors
-    if let ApiError::InternalServerError(_) = api_error.as_ref() {
-        error!("Error processing HTTP request: {api_error:?}");
-    } else {
-        error!("Error processing HTTP request: {api_error:#}");
-    }
-
-    api_error.into_response()
+    tracing::error!("Error processing HTTP request: {:?}", err);
+    err.downcast::<ApiError>()
+        .expect("handler should always return api error")
+        .into_response()
 }

libs/utils/src/http/json.rs

@@ -1,9 +1,7 @@
-use std::fmt::Display;
-
 use anyhow::Context;
 use bytes::Buf;
 use hyper::{header, Body, Request, Response, StatusCode};
-use serde::{Deserialize, Serialize, Serializer};
+use serde::{Deserialize, Serialize};
 
 use super::error::ApiError;
 
@@ -33,12 +31,3 @@ pub fn json_response<T: Serialize>(
         .map_err(|e| ApiError::InternalServerError(e.into()))?;
     Ok(response)
 }
-
-/// Serialize through Display trait.
-pub fn display_serialize<S, F>(z: &F, s: S) -> Result<S::Ok, S::Error>
-where
-    S: Serializer,
-    F: Display,
-{
-    s.serialize_str(&format!("{}", z))
-}