feat(acceptance): 增加生产同构媒体压力验收模式

引入动态流量门禁、隔离验收身份与协议级 Gemini/Volces 模拟器，覆盖双站点 API、Worker、PostgreSQL、River、账务、回调和媒体物化链路。

新增 P24/P28/P32 容量阶梯、Worker 强杀恢复、真实小流量 canary、CAS 放量和失败保持 validation 的生产编排；Worker 执行槽、连接池、媒体并发和双站点副本数改为环境配置。

验证：Go 全量测试、真实 PostgreSQL 迁移集成测试、迁移安全检查、OpenAPI 生成、ShellCheck、Kustomize、gofmt 和 git diff --check。

deploy/kubernetes/acceptance/kustomization.yaml

@@ -0,0 +1,4 @@
+apiVersion: kustomize.config.k8s.io/v1beta1
+kind: Kustomization
+resources:
+  - protocol-emulator.yaml

deploy/kubernetes/acceptance/protocol-emulator.yaml

@@ -0,0 +1,67 @@
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: easyai-acceptance-emulator
+  namespace: easyai
+  labels:
+    app.kubernetes.io/name: easyai-acceptance-emulator
+    app.kubernetes.io/part-of: easyai-ai-gateway
+spec:
+  replicas: 1
+  selector:
+    matchLabels:
+      app.kubernetes.io/name: easyai-acceptance-emulator
+  template:
+    metadata:
+      labels:
+        app.kubernetes.io/name: easyai-acceptance-emulator
+        app.kubernetes.io/part-of: easyai-ai-gateway
+    spec:
+      nodeSelector:
+        easyai.io/site: ningbo
+        easyai.io/workload: "true"
+      securityContext:
+        runAsNonRoot: true
+        runAsUser: 10001
+        runAsGroup: 10001
+        seccompProfile:
+          type: RuntimeDefault
+      containers:
+        - name: emulator
+          image: easyai-api
+          command: ["/app/easyai-ai-gateway-acceptance-emulator"]
+          env:
+            - name: HTTP_ADDR
+              value: ":8090"
+          ports:
+            - name: http
+              containerPort: 8090
+          readinessProbe:
+            httpGet:
+              path: /healthz
+              port: http
+          resources:
+            requests:
+              cpu: 100m
+              memory: 128Mi
+            limits:
+              cpu: "2"
+              memory: 1Gi
+          securityContext:
+            allowPrivilegeEscalation: false
+            capabilities:
+              drop: ["ALL"]
+            readOnlyRootFilesystem: true
+---
+apiVersion: v1
+kind: Service
+metadata:
+  name: easyai-acceptance-emulator
+  namespace: easyai
+spec:
+  selector:
+    app.kubernetes.io/name: easyai-acceptance-emulator
+  ports:
+    - name: http
+      port: 8090
+      targetPort: http

deploy/kubernetes/easyai-ai-gateway-cluster-release

@@ -14,10 +14,40 @@ source "$config_file"
 : "${NAMESPACE:=easyai}"
 : "${PUBLIC_BASE_URL:=https://ai.51easyai.com}"
 : "${K3S_BIN:=/usr/local/bin/k3s}"
+: "${AI_GATEWAY_WORKER_REPLICAS_NINGBO:=1}"
+: "${AI_GATEWAY_WORKER_REPLICAS_HONGKONG:=1}"
+: "${AI_GATEWAY_ASYNC_WORKER_INSTANCE_HARD_LIMIT:=24}"
+: "${AI_GATEWAY_ASYNC_WORKER_HARD_LIMIT:=48}"
+: "${AI_GATEWAY_DATABASE_MAX_CONNS:=32}"
+: "${AI_GATEWAY_MEDIA_MATERIALIZATION_CONCURRENCY:=24}"
+: "${AI_GATEWAY_MEDIA_REQUEST_CONCURRENCY:=24}"
 [[ $RELEASES_DIR == /* && $NAMESPACE =~ ^[a-z0-9-]+$ ]] || {
   echo 'invalid cluster release configuration' >&2
   exit 1
 }
+for capacity_value in \
+  "$AI_GATEWAY_WORKER_REPLICAS_NINGBO" \
+  "$AI_GATEWAY_WORKER_REPLICAS_HONGKONG" \
+  "$AI_GATEWAY_ASYNC_WORKER_INSTANCE_HARD_LIMIT" \
+  "$AI_GATEWAY_ASYNC_WORKER_HARD_LIMIT" \
+  "$AI_GATEWAY_DATABASE_MAX_CONNS" \
+  "$AI_GATEWAY_MEDIA_MATERIALIZATION_CONCURRENCY" \
+  "$AI_GATEWAY_MEDIA_REQUEST_CONCURRENCY"; do
+  [[ $capacity_value =~ ^[1-9][0-9]*$ ]] || {
+    echo 'worker capacity configuration must contain positive integers' >&2
+    exit 1
+  }
+done
+(( AI_GATEWAY_WORKER_REPLICAS_NINGBO <= 16 &&
+   AI_GATEWAY_WORKER_REPLICAS_HONGKONG <= 16 &&
+   AI_GATEWAY_ASYNC_WORKER_INSTANCE_HARD_LIMIT <= 256 &&
+   AI_GATEWAY_ASYNC_WORKER_HARD_LIMIT <= 512 &&
+   AI_GATEWAY_DATABASE_MAX_CONNS <= 256 &&
+   AI_GATEWAY_MEDIA_MATERIALIZATION_CONCURRENCY <= 256 &&
+   AI_GATEWAY_MEDIA_REQUEST_CONCURRENCY <= 1024 )) || {
+  echo 'worker capacity configuration exceeds the release safety bounds' >&2
+  exit 1
+}
 [[ $PUBLIC_BASE_URL =~ ^https://[A-Za-z0-9.-]+$ ]] || {
   echo 'invalid public base URL' >&2
   exit 1
@@ -87,11 +117,47 @@ verify_site() {
 rollout_worker_site() {
   local site=$1
   local api_image=$2
-  "${kubectl[@]}" set image "deployment/easyai-worker-$site" -n "$NAMESPACE" \
-    "worker=$api_image"
+  local replicas
+  case $site in
+    ningbo) replicas=$AI_GATEWAY_WORKER_REPLICAS_NINGBO ;;
+    hongkong) replicas=$AI_GATEWAY_WORKER_REPLICAS_HONGKONG ;;
+    *) return 1 ;;
+  esac
+  "${kubectl[@]}" set env "deployment/easyai-worker-$site" -n "$NAMESPACE" \
+    "AI_GATEWAY_ASYNC_WORKER_INSTANCE_HARD_LIMIT=$AI_GATEWAY_ASYNC_WORKER_INSTANCE_HARD_LIMIT" \
+    "AI_GATEWAY_ASYNC_WORKER_HARD_LIMIT=$AI_GATEWAY_ASYNC_WORKER_HARD_LIMIT" \
+    "AI_GATEWAY_DATABASE_MAX_CONNS=$AI_GATEWAY_DATABASE_MAX_CONNS" \
+    "AI_GATEWAY_MEDIA_MATERIALIZATION_CONCURRENCY=$AI_GATEWAY_MEDIA_MATERIALIZATION_CONCURRENCY" \
+    "AI_GATEWAY_MEDIA_REQUEST_CONCURRENCY=$AI_GATEWAY_MEDIA_REQUEST_CONCURRENCY"
+  "${kubectl[@]}" scale "deployment/easyai-worker-$site" -n "$NAMESPACE" \
+    --replicas="$replicas"
+  if [[ -n $api_image ]]; then
+    "${kubectl[@]}" set image "deployment/easyai-worker-$site" -n "$NAMESPACE" \
+      "worker=$api_image"
+  fi
   "${kubectl[@]}" rollout status "deployment/easyai-worker-$site" -n "$NAMESPACE" --timeout=300s
   [[ $("${kubectl[@]}" get deployment "easyai-worker-$site" -n "$NAMESPACE" \
     -o jsonpath='{.spec.template.spec.containers[?(@.name=="worker")].env[?(@.name=="AI_GATEWAY_PROCESS_ROLE")].value}') == worker ]]
+  [[ $("${kubectl[@]}" get deployment "easyai-worker-$site" -n "$NAMESPACE" \
+    -o jsonpath='{.status.readyReplicas}') == "$replicas" ]]
+}
+
+rollout_api_capacity_site() {
+  local site=$1
+  "${kubectl[@]}" set env "deployment/easyai-api-$site" -n "$NAMESPACE" \
+    "AI_GATEWAY_ASYNC_WORKER_HARD_LIMIT=$AI_GATEWAY_ASYNC_WORKER_HARD_LIMIT"
+  "${kubectl[@]}" rollout status "deployment/easyai-api-$site" -n "$NAMESPACE" --timeout=300s
+}
+
+apply_capacity_config() {
+  rollout_worker_site hongkong ""
+  rollout_worker_site ningbo ""
+  rollout_api_capacity_site hongkong
+  rollout_api_capacity_site ningbo
+  verify_site hongkong
+  verify_site ningbo
+  wait_for_url 'public readiness' "$PUBLIC_BASE_URL/api/v1/readyz" '"ok":true'
+  echo "production_capacity=PASS ningbo_replicas=$AI_GATEWAY_WORKER_REPLICAS_NINGBO hongkong_replicas=$AI_GATEWAY_WORKER_REPLICAS_HONGKONG instance_limit=$AI_GATEWAY_ASYNC_WORKER_INSTANCE_HARD_LIMIT global_limit=$AI_GATEWAY_ASYNC_WORKER_HARD_LIMIT database_pool=$AI_GATEWAY_DATABASE_MAX_CONNS media_concurrency=$AI_GATEWAY_MEDIA_MATERIALIZATION_CONCURRENCY"
 }
 
 rollout_site() {
@@ -101,6 +167,8 @@ rollout_site() {
   local api_changed=$4
   local web_changed=$5
   if [[ $api_changed == true ]]; then
+    "${kubectl[@]}" set env "deployment/easyai-api-$site" -n "$NAMESPACE" \
+      "AI_GATEWAY_ASYNC_WORKER_HARD_LIMIT=$AI_GATEWAY_ASYNC_WORKER_HARD_LIMIT"
     "${kubectl[@]}" set image "deployment/easyai-api-$site" -n "$NAMESPACE" \
       "api=$api_image"
     "${kubectl[@]}" rollout status "deployment/easyai-api-$site" -n "$NAMESPACE" --timeout=300s
@@ -305,8 +373,12 @@ case ${1:-} in
     [[ -f $target && ! -L $target ]] || { echo 'unknown historical release' >&2; exit 1; }
     activate_manifest "$target" rollback
     ;;
+  capacity)
+    [[ $# -eq 1 ]] || exit 64
+    apply_capacity_config
+    ;;
   *)
-    echo 'usage: easyai-ai-gateway-cluster-release {status|adopt <manifest>|deploy <manifest>|rollback <sha>}' >&2
+    echo 'usage: easyai-ai-gateway-cluster-release {status|adopt <manifest>|deploy <manifest>|rollback <sha>|capacity}' >&2
     exit 64
     ;;
 esac

deploy/kubernetes/easyai-ai-gateway-cluster-release.conf.example

@@ -2,3 +2,13 @@ RELEASES_DIR=/var/lib/easyai-ai-gateway-cluster-release/releases
 NAMESPACE=easyai
 PUBLIC_BASE_URL=https://ai.51easyai.com
 K3S_BIN=/usr/local/bin/k3s
+
+# Worker 容量只需修改本发布配置并执行获授权的 `capacity`，不再修改 Go 代码。
+# 副本数属于 Deployment 配置，不能由容器内部环境变量改变；发布工具读取这两个变量并写入 replicas。
+AI_GATEWAY_WORKER_REPLICAS_NINGBO=1
+AI_GATEWAY_WORKER_REPLICAS_HONGKONG=1
+AI_GATEWAY_ASYNC_WORKER_INSTANCE_HARD_LIMIT=24
+AI_GATEWAY_ASYNC_WORKER_HARD_LIMIT=48
+AI_GATEWAY_DATABASE_MAX_CONNS=32
+AI_GATEWAY_MEDIA_MATERIALIZATION_CONCURRENCY=24
+AI_GATEWAY_MEDIA_REQUEST_CONCURRENCY=24

deploy/kubernetes/production/application-config.yaml

@@ -37,5 +37,5 @@ data:
   AI_GATEWAY_LOCAL_RESULT_MAX_BYTES: "268435456"
   AI_GATEWAY_LOCAL_RESULT_MAX_TASK_BYTES: "536870912"
   AI_GATEWAY_ASYNC_QUEUE_WORKER_ENABLED: "true"
-  AI_GATEWAY_ASYNC_WORKER_HARD_LIMIT: "2048"
+  AI_GATEWAY_ASYNC_WORKER_HARD_LIMIT: "48"
   AI_GATEWAY_ASYNC_WORKER_REFRESH_INTERVAL_SECONDS: "5"

deploy/kubernetes/production/application.yaml

@@ -316,6 +316,8 @@ spec:
               value: "24"
             - name: AI_GATEWAY_MEDIA_MATERIALIZATION_CONCURRENCY
               value: "24"
+            - name: AI_GATEWAY_MEDIA_REQUEST_CONCURRENCY
+              value: "24"
             - name: POD_NAMESPACE
               valueFrom:
                 fieldRef:
@@ -445,6 +447,8 @@ spec:
               value: "24"
             - name: AI_GATEWAY_MEDIA_MATERIALIZATION_CONCURRENCY
               value: "24"
+            - name: AI_GATEWAY_MEDIA_REQUEST_CONCURRENCY
+              value: "24"
             - name: POD_NAMESPACE
               valueFrom:
                 fieldRef: