AztecProtocol · PhilWindle · Nov 28, 2024 · Nov 18, 2024 · Nov 19, 2024 · Nov 26, 2024
diff --git a/docker-compose.provernet.yml b/docker-compose.provernet.yml
@@ -65,7 +65,8 @@ services:
       ARCHIVER_POLLING_INTERVAL_MS: 1000
       ARCHIVER_VIEM_POLLING_INTERVAL_MS: 1000
       PROVER_VIEM_POLLING_INTERVAL_MS: 1000
-      PROVER_AGENT_ENABLED: false
+      PROVER_AGENT_COUNT: 0
+      PROVER_BROKER_HOST: http://aztec-prover-broker
       PROVER_PUBLISHER_PRIVATE_KEY: "0xdbda1821b80551c9d65939329250298aa3472ba22feea921c0cf5d620ea67b97"
       PROVER_REAL_PROOFS: "${PROVER_REAL_PROOFS:-false}"
       PROVER_MINIMUM_ESCROW_AMOUNT: 1000000000
@@ -76,6 +77,8 @@ services:
     depends_on:
       aztec-node:
         condition: service_healthy
+      aztec-prover-broker:
+        condition: service_healthy
     healthcheck:
       test: [ "CMD", "curl", "-fSs", "http://127.0.0.1:80/status" ]
       interval: 3s
@@ -84,6 +87,21 @@ services:
     command: [ "start", "--prover-node", "--archiver" ]
     restart: on-failure:5
 
+  aztec-prover-broker:
+    image: "aztecprotocol/${IMAGE:-aztec:master}"
+    ports:
+      - "8084:80"
+    environment:
+      LOG_LEVEL: verbose
+      AZTEC_PORT: 80
+    healthcheck:
+      test: [ "CMD", "curl", "-fSs", "http://127.0.0.1:80/status" ]
+      interval: 3s
+      timeout: 30s
+      start_period: 120s
+    command: [ "start", "--prover-broker" ]
+    restart: on-failure:5
+
   # Prover agent that connects to the prover-node for fetching proving jobs and executing them
   # Multiple instances can be run, or PROVER_AGENT_CONCURRENCY can be increased to run multiple workers in a single instance
   aztec-prover-agent:
@@ -93,23 +111,24 @@ services:
     environment:
       LOG_LEVEL: verbose
       ETHEREUM_HOST: http://ethereum:8545
-      AZTEC_NODE_URL: http://aztec-prover # Deprecated, use PROVER_JOB_SOURCE_URL
-      PROVER_JOB_SOURCE_URL: http://aztec-prover
+      PROVER_BROKER_HOST: http://aztec-prover-broker
       L1_CHAIN_ID: 31337
       AZTEC_PORT: 80
       PROVER_REAL_PROOFS: "${PROVER_REAL_PROOFS:-false}"
       PROVER_TEST_DELAY_MS: "${PROVER_TEST_DELAY_MS:-0}"
-      PROVER_AGENT_CONCURRENCY: 2
       BB_SKIP_CLEANUP: "${BB_SKIP_CLEANUP:-0}" # Persist tmp dirs for debugging
       PROVER_ID: "${PROVER_ID:-0x01}"
     volumes:
       - ./log/aztec-prover-agent/:/usr/src/yarn-project/aztec/log:rw
       - ./cache/bb-crs/:/root/.bb-crs:rw
       - ./workdir/bb-prover/:/usr/src/yarn-project/bb:rw
     depends_on:
-      aztec-prover:
+      aztec-prover-broker:
         condition: service_healthy
-    command: [ "start", "--prover" ]
+    command: [ "start", "--prover-agent" ]
+    deploy:
+      mode: replicated
+      replicas: 2
     restart: on-failure:5
     healthcheck:
       test: [ "CMD", "curl", "-fSs", "http://127.0.0.1:80/status" ]

diff --git a/spartan/aztec-network/files/config/setup-service-addresses.sh b/spartan/aztec-network/files/config/setup-service-addresses.sh
@@ -79,10 +79,19 @@ else
     PROVER_NODE_ADDR="http://${SERVICE_NAME}-prover-node.${NAMESPACE}:${PROVER_NODE_PORT}"
 fi
 
+if [ "${PROVER_BROKER_EXTERNAL_HOST}" != "" ]; then
+    PROVER_BROKER_ADDR="${PROVER_BROKER_EXTERNAL_HOST}"
+elif [ "${NETWORK_PUBLIC}" = "true" ]; then
+    PROVER_BROKER_ADDR=$(get_service_address "prover-broker" "${PROVER_BROKER_PORT}")
+else
+    PROVER_BROKER_ADDR="http://${SERVICE_NAME}-prover-broker.${NAMESPACE}:${PROVER_BROKER_PORT}"
+fi
+
 
 # Write addresses to file for sourcing
 echo "export ETHEREUM_HOST=${ETHEREUM_ADDR}" >> /shared/config/service-addresses
 echo "export BOOT_NODE_HOST=${BOOT_NODE_ADDR}" >> /shared/config/service-addresses
 echo "export PROVER_NODE_HOST=${PROVER_NODE_ADDR}" >> /shared/config/service-addresses
+echo "export PROVER_BROKER_HOST=${PROVER_BROKER_ADDR}" >> /shared/config/service-addresses
 echo "Addresses configured:"
 cat /shared/config/service-addresses
diff --git a/spartan/aztec-network/templates/_helpers.tpl b/spartan/aztec-network/templates/_helpers.tpl
@@ -165,6 +165,8 @@ Service Address Setup Container
       value: "{{ .Values.proverNode.externalHost }}"
     - name: PROVER_NODE_PORT
       value: "{{ .Values.proverNode.service.nodePort }}"
+    - name: PROVER_BROKER_PORT
+      value: "{{ .Values.proverBroker.service.nodePort }}"
     - name: SERVICE_NAME
       value: {{ include "aztec-network.fullname" . }}
   volumeMounts:

diff --git a/spartan/aztec-network/templates/prover-agent.yaml b/spartan/aztec-network/templates/prover-agent.yaml
@@ -50,11 +50,11 @@ spec:
             - -c
             - |
               source /shared/config/service-addresses
-              until curl -s -X POST ${PROVER_NODE_HOST}/status; do
-                echo "Waiting for Prover node ${PROVER_NODE_HOST} ..."
+              until curl -s -X POST ${PROVER_BROKER_HOST}/status; do
+                echo "Waiting for broker ${PROVER_BROKER_HOST} ..."
                 sleep 5
               done
-              echo "Prover node is ready!"
+              echo "Broker is ready!"
               {{- if .Values.telemetry.enabled }}
               until curl --head --silent {{ include "aztec-network.otelCollectorMetricsEndpoint" . }} > /dev/null; do
                 echo "Waiting for OpenTelemetry collector..."
@@ -77,8 +77,7 @@ spec:
             - "-c"
             - |
               source /shared/config/service-addresses && \
-              PROVER_JOB_SOURCE_URL=${PROVER_NODE_HOST} \
-              node --no-warnings /usr/src/yarn-project/aztec/dest/bin/index.js start --prover
+              node --no-warnings /usr/src/yarn-project/aztec/dest/bin/index.js start --prover-agent
           env:
             - name: AZTEC_PORT
               value: "{{ .Values.proverAgent.service.nodePort }}"
@@ -90,12 +89,12 @@ spec:
               value: "{{ .Values.proverAgent.debug }}"
             - name: PROVER_REAL_PROOFS
               value: "{{ .Values.proverAgent.realProofs }}"
-            - name: PROVER_AGENT_ENABLED
-              value: "true"
-            - name: PROVER_AGENT_CONCURRENCY
-              value: {{ .Values.proverAgent.concurrency | quote }}
-            - name: HARDWARE_CONCURRENCY
-              value: {{ .Values.proverAgent.bb.hardwareConcurrency | quote }}
+            - name: PROVER_AGENT_COUNT
+              value: "1"
+            - name: PROVER_AGENT_POLL_INTERVAL_MS
+              value: "{{ .Values.proverAgent.pollIntervalMs }}"
+            - name: PROVER_AGENT_PROOF_TYPES
+              value: {{ join "," .Values.proverAgent.proofTypes | quote }}
             - name: OTEL_RESOURCE_ATTRIBUTES
               value: service.name={{ .Release.Name }},service.namespace={{ .Release.Namespace }},service.version={{ .Chart.AppVersion }},environment={{ .Values.environment | default "production" }}
             - name: OTEL_EXPORTER_OTLP_METRICS_ENDPOINT

diff --git a/spartan/aztec-network/templates/prover-broker.yaml b/spartan/aztec-network/templates/prover-broker.yaml
@@ -0,0 +1,104 @@
+{{- if .Values.proverBroker.enabled }}
+apiVersion: apps/v1
+kind: ReplicaSet
+metadata:
+  name: {{ include "aztec-network.fullname" . }}-prover-broker
+  labels:
+    {{- include "aztec-network.labels" . | nindent 4 }}
+spec:
+  replicas: {{ .Values.proverBroker.replicas }}
+  selector:
+    matchLabels:
+      {{- include "aztec-network.selectorLabels" . | nindent 6 }}
+      app: prover-broker
+  template:
+    metadata:
+      labels:
+        {{- include "aztec-network.selectorLabels" . | nindent 8 }}
+        app: prover-broker
+    spec:
+      serviceAccountName: {{ include "aztec-network.fullname" . }}-node
+      {{- if .Values.network.public }}
+      hostNetwork: true
+      {{- end }}
+      volumes:
+        - name: config
+          emptyDir: {}
+        - name: scripts
+          configMap:
+            name: {{ include "aztec-network.fullname" . }}-scripts
+      initContainers:
+        {{- include "aztec-network.serviceAddressSetupContainer" . | nindent 8 }}
+        - name: wait-for-prover-node
+          image: {{ .Values.images.aztec.image }}
+          command:
+            - /bin/bash
+            - -c
+            - |
+              source /shared/config/service-addresses
+              {{- if .Values.telemetry.enabled }}
+              until curl --head --silent {{ include "aztec-network.otelCollectorMetricsEndpoint" . }} > /dev/null; do
+                echo "Waiting for OpenTelemetry collector..."
+                sleep 5
+              done
+              echo "OpenTelemetry collector is ready!"
+              {{- end }}
+          volumeMounts:
+            - name: config
+              mountPath: /shared/config
+      containers:
+        - name: prover-broker
+          image: "{{ .Values.images.aztec.image }}"
+          imagePullPolicy: {{ .Values.images.aztec.pullPolicy }}
+          volumeMounts:
+            - name: config
+              mountPath: /shared/config
+          command:
+            - "/bin/bash"
+            - "-c"
+            - |
+              source /shared/config/service-addresses && \
+              node --no-warnings /usr/src/yarn-project/aztec/dest/bin/index.js start --prover-broker
+          env:
+            - name: AZTEC_PORT
+              value: "{{ .Values.proverBroker.service.nodePort }}"
+            - name: LOG_LEVEL
+              value: "{{ .Values.proverBroker.logLevel }}"
+            - name: LOG_JSON
+              value: "1"
+            - name: DEBUG
+              value: "{{ .Values.proverBroker.debug }}"
+            - name: PROVER_BROKER_POLL_INTERVAL_MS
+              value: "{{ .Values.proverBroker.pollIntervalMs }}"
+            - name: PROVER_BROKER_JOB_TIMEOUT_MS
+              value: "{{ .Values.proverBroker.jobTimeoutMs }}"
+            - name: PROVER_BROKER_JOB_MAX_RETRIES
+              value: "{{ .Values.proverBroker.jobMaxRetries }}"
+            - name: PROVER_BROKER_DATA_DIRECTORY
+              value: "{{ .Values.proverBroker.dataDirectory }}"
+            - name: OTEL_RESOURCE_ATTRIBUTES
+              value: service.name={{ .Release.Name }},service.namespace={{ .Release.Namespace }},service.version={{ .Chart.AppVersion }},environment={{ .Values.environment | default "production" }}
+            - name: OTEL_EXPORTER_OTLP_METRICS_ENDPOINT
+              value: {{ include "aztec-network.otelCollectorMetricsEndpoint" . | quote }}
+            - name: OTEL_EXPORTER_OTLP_TRACES_ENDPOINT
+              value: {{ include "aztec-network.otelCollectorTracesEndpoint" . | quote }}
+            - name: OTEL_EXPORTER_OTLP_LOGS_ENDPOINT
+              value: {{ include "aztec-network.otelCollectorLogsEndpoint" . | quote }}
+          resources:
+            {{- toYaml .Values.proverBroker.resources | nindent 12 }}
+---
+apiVersion: v1
+kind: Service
+metadata:
+  name: {{ include "aztec-network.fullname" . }}-prover-broker
+  labels:
+    {{- include "aztec-network.labels" . | nindent 4 }}
+spec:
+  type: ClusterIP
+  selector:
+    {{- include "aztec-network.selectorLabels" . | nindent 4 }}
+    app: prover-broker
+  ports:
+    - port: {{ .Values.proverBroker.service.nodePort }}
+      name: node
+{{ end }}
diff --git a/spartan/aztec-network/templates/prover-node.yaml b/spartan/aztec-network/templates/prover-node.yaml
@@ -39,6 +39,17 @@ spec:
                 sleep 5
               done
               echo "Ethereum node is ready!"
+
+              if [ "${PROVER_BROKER_ENABLED}" == "false" ]; then
+                until curl -s -X POST ${PROVER_BROKER_HOST}/status; do
+                  echo "Waiting for broker ${PROVER_BROKER_HOST} ..."
+                  sleep 5
+                done
+                echo "Broker is ready!"
+              else
+                echo "Using built-in job broker"
+              fi
+
               {{- if .Values.telemetry.enabled }}
               until curl --head --silent {{ include "aztec-network.otelCollectorMetricsEndpoint" . }} > /dev/null; do
                 echo "Waiting for OpenTelemetry collector..."
@@ -54,6 +65,10 @@ spec:
           volumeMounts:
             - name: config
               mountPath: /shared/config
+          env:
+            - name: PROVER_BROKER_ENABLED
+              value: "{{ .Values.proverNode.proverBroker.enabled }}"
+
         - name: configure-prover-env
           image: "{{ .Values.images.aztec.image }}"
           imagePullPolicy: {{ .Values.images.aztec.pullPolicy }}
@@ -107,15 +122,26 @@ spec:
               value: "{{ .Values.proverNode.debug }}"
             - name: PROVER_REAL_PROOFS
               value: "{{ .Values.proverNode.realProofs }}"
-            - name: PROVER_AGENT_ENABLED
-              value: "{{ .Values.proverNode.proverAgentEnabled }}"
+            - name: PROVER_AGENT_COUNT
+              value: "{{ .Values.proverNode.proverAgent.count }}"
+            - name: PROVER_AGENT_POLL_INTERVAL_MS
+              value: "{{ .Values.proverNode.proverAgent.pollIntervalMs }}"
+            - name: PROVER_AGENT_PROOF_TYPES
+              value: {{ join "," .Values.proverNode.proverAgent.proofTypes | quote }}
+            - name: PROVER_BROKER_ENABLED
+              value: "{{ .Values.proverNode.proverBroker.enabled }}"
+            - name: PROVER_BROKER_POLL_INTERVAL_MS
+              value: "{{ .Values.proverNode.proverBroker.pollIntervalMs }}"
+            - name: PROVER_BROKER_JOB_TIMEOUT_MS
+              value: "{{ .Values.proverNode.proverBroker.jobTimeoutMs }}"
+            - name: PROVER_BROKER_JOB_MAX_RETRIES
+              value: "{{ .Values.proverNode.proverBroker.jobMaxRetries }}"
+            - name: PROVER_BROKER_DATA_DIRECTORY
+              value: "{{ .Values.proverNode.proverBroker.dataDirectory }}"
             - name: PROVER_PUBLISHER_PRIVATE_KEY
               value: "0xac0974bec39a17e36ba4a6b4d238ff944bacb478cbed5efcae784d7bf4f2ff80"
             - name: OTEL_RESOURCE_ATTRIBUTES
               value: service.name={{ .Release.Name }},service.namespace={{ .Release.Namespace }},service.version={{ .Chart.AppVersion }},environment={{ .Values.environment | default "production" }}
-            # get private proofs from the boot node
-            - name: PROVER_JOB_SOURCE_URL
-              value: "http://$(POD_IP):{{ .Values.proverNode.service.nodePort }}"
             - name: OTEL_EXPORTER_OTLP_METRICS_ENDPOINT
               value: {{ include "aztec-network.otelCollectorMetricsEndpoint" . | quote }}
             - name: OTEL_EXPORTER_OTLP_TRACES_ENDPOINT

diff --git a/spartan/aztec-network/values.yaml b/spartan/aztec-network/values.yaml
@@ -118,7 +118,16 @@ proverNode:
   logLevel: "debug"
   debug: "aztec:*,-aztec:avm_simulator*,-aztec:libp2p_service*,-aztec:circuits:artifact_hash,-json-rpc*,-aztec:world-state:database,-aztec:l2_block_stream*"
   realProofs: false
-  proverAgentEnabled: false
+  proverAgent:
+    count: 0
+    pollIntervalMs: 1000
+    proofTypes: []
+  proverBroker:
+    enabled: false
+    jobTimeoutMs: 30000
+    pollIntervalMs: 1000
+    jobMaxRetries: 3
+    dataDirectory: ""
   resources:
     requests:
       memory: "2Gi"
@@ -206,17 +215,32 @@ proverAgent:
     nodePort: 8083
   enabled: true
   replicas: 1
+  pollIntervalMs: 1000
+  proofTypes: ["foo", "bar", "baz"]
   gke:
     spotEnabled: false
   logLevel: "debug"
   debug: "aztec:*,-aztec:avm_simulator*,-aztec:libp2p_service*,-aztec:circuits:artifact_hash,-json-rpc*,-aztec:world-state:database,-aztec:l2_block_stream*"
   realProofs: false
-  concurrency: 1
   bb:
     hardwareConcurrency: ""
   nodeSelector: {}
   resources: {}
 
+proverBroker:
+  service:
+    nodePort: 8084
+  enabled: true
+  replicas: 1
+  jobTimeoutMs: 30000
+  pollIntervalMs: 1000
+  jobMaxRetries: 3
+  dataDirectory: ""
+  logLevel: "debug"
+  debug: "aztec:*,-aztec:avm_simulator*,-aztec:libp2p_service*,-aztec:circuits:artifact_hash,-json-rpc*,-aztec:world-state:database,-aztec:l2_block_stream*"
+  nodeSelector: {}
+  resources: {}
+
 jobs:
   deployL1Verifier:
     enable: false
diff --git a/yarn-project/aztec-node/src/aztec-node/server.ts b/yarn-project/aztec-node/src/aztec-node/server.ts
@@ -165,6 +165,7 @@ export class AztecNodeService implements AztecNode {
     // now create the merkle trees and the world state synchronizer
     const worldStateSynchronizer = await createWorldStateSynchronizer(config, archiver, telemetry);
     const proofVerifier = config.realProofs ? await BBCircuitVerifier.new(config) : new TestCircuitVerifier();
+    log.info(`Aztec node accepting ${config.realProofs ? 'real' : 'test'} proofs`);
 
     // create the tx pool and the p2p client, which will need the l2 block source
     const p2pClient = await createP2PClient(config, archiver, proofVerifier, worldStateSynchronizer, telemetry);