feat: complete production blue-green deployment

deploy/scripts/deploy-blue-green-service.sh

@@ -33,7 +33,7 @@ env_value() {
 
 service_contract() {
   case "$SERVICE:$COMPOSE_FILE" in
-    gateway:*) echo "gateway|9000|9000|gateway|gateway" ;;
+    gateway:*) echo "gateway|9000|9000|gateway|gateway-active" ;;
     auth:*) echo "auth|9100|9100|auth-service|auth" ;;
     iam:*) echo "iam|9201|9201|iam-service|iam" ;;
     firstleg:*) echo "business|9202|9202|firstleg-service|firstleg" ;;
@@ -44,11 +44,11 @@ service_contract() {
     print:*) echo "print|9207|9207|print-service|print" ;;
     integration-service:*) echo "integration|9210|9210|integration-service|integration-service" ;;
     audit-service:*) echo "audit|9700|9700|audit-service|audit-service" ;;
-    gateway-overseas:*) echo "gateway-overseas|9300|9300|gateway-overseas|gateway-overseas" ;;
+    gateway-overseas:*) echo "gateway-overseas|9300|9300|gateway-overseas|gateway-overseas-active" ;;
     lastleg-service:*) echo "lastleg|9303|9303|lastleg-service|lastleg-service" ;;
     warehouse-overseas-service:*) echo "warehouse-overseas|9305|9305|warehouse-overseas-service|warehouse-overseas-service" ;;
-    event-bridge-service:*overseas*) echo "event-bridge-overseas|9312|9313|event-bridge-service|event-bridge-overseas" ;;
+    event-bridge-service:*overseas*) echo "event-bridge-overseas|9312|9313|event-bridge-overseas|event-bridge-overseas" ;;
-    event-bridge-service:*) echo "event-bridge|9312|9312|event-bridge-service|event-bridge-cn" ;;
+    event-bridge-service:*) echo "event-bridge|9312|9312|event-bridge-cn|event-bridge-cn" ;;
     *) echo "不支持蓝绿部署的服务: $SERVICE" >&2; exit 2 ;;
   esac
 }
@@ -173,6 +173,7 @@ start_worker_candidate() {
       -e "BAIYUE_RELEASE_ID=$IMAGE_TAG" "$SERVICE" >/dev/null; then
     return 1
   fi
+  docker network connect "$ROUTING_NETWORK" "$worker_candidate"
   if ! wait_for_container "$worker_candidate" "候选 Worker"; then
     docker logs --tail 200 "$worker_candidate" 2>&1 || true
     docker rm -f "$worker_candidate" >/dev/null 2>&1 || true
@@ -182,38 +183,50 @@ start_worker_candidate() {
   [ "$worker_image" = "$IMAGE" ] || return 1
 }
 
-connect_routing_alias() {
+connect_routing_aliases() {
-  alias_name="$1"
+  container="$1"
-  container="$2"
+  shift
   docker network inspect "$ROUTING_NETWORK" --format '{{json .Containers}}' | grep -q "\"$container\"" \
     && return 0
-  docker network connect --alias "$alias_name" "$ROUTING_NETWORK" "$container"
+  set -- docker network connect $(for alias_name in "$@"; do printf '%s\n' --alias "$alias_name"; done)
+  "$@" "$ROUTING_NETWORK" "$container"
 }
 
 bootstrap_legacy_routes() {
   for pair in \
-    gateway:gateway auth:auth iam:iam firstleg:business finance:finance order:order \
+    gateway-active,gateway:gateway auth,auth-service:auth iam,iam-service:iam \
-    warehouse:warehouse file:file print:print integration-service:integration audit-service:audit \
+    firstleg,firstleg-service:business finance,finance-service:finance order,order-service:order \
-    gateway-overseas:gateway-overseas lastleg-service:lastleg \
+    warehouse,warehouse-service:warehouse file,file-service:file print,print-service:print \
-    warehouse-overseas-service:warehouse-overseas event-bridge-cn:event-bridge \
+    integration-service:integration audit-service:audit \
-    event-bridge-overseas:event-bridge-overseas
+    gateway-overseas-active,gateway-overseas:gateway-overseas \
+    lastleg-service:lastleg warehouse-overseas-service:warehouse-overseas \
+    event-bridge-cn:event-bridge event-bridge-overseas:event-bridge-overseas
   do
-    alias_name="${pair%%:*}"
+    aliases="${pair%%:*}"
     stem="${pair#*:}"
     container="firstleg-${DEPLOY_ENV}-${stem}"
     if docker inspect "$container" >/dev/null 2>&1; then
-      connect_routing_alias "$alias_name" "$container" || true
+      connect_routing_aliases "$container" $(printf '%s' "$aliases" | tr ',' ' ') || true
     fi
   done
   for container in firstleg-prod-admin-web firstleg-prod-client-web; do
     if docker inspect "$container" >/dev/null 2>&1; then
-      connect_routing_alias "${container##*-}" "$container" || true
+      connect_routing_aliases "$container" "${container##*-}" || true
     fi
   done
 }
 
 switch_gateway_proxy() {
   [ "$SERVICE" = "gateway" ] || [ "$SERVICE" = "gateway-overseas" ] || return 0
+  if [ "$SERVICE" = "gateway" ]; then required_alias=gateway-active; else required_alias=gateway-overseas-active; fi
+  for frontend in firstleg-prod-admin-web-blue firstleg-prod-admin-web-green \
+    firstleg-prod-client-web-blue firstleg-prod-client-web-green; do
+    if docker inspect "$frontend" >/dev/null 2>&1 \
+      && ! docker exec "$frontend" nginx -T 2>&1 | grep -q "$required_alias"; then
+      echo "$frontend 尚未使用稳定上游 $required_alias，拒绝切换 Gateway" >&2
+      return 1
+    fi
+  done
   if [ "$SERVICE" = "gateway" ]; then
     config="$(env_value GATEWAY_BLUE_GREEN_NGINX_CONFIG)"
     config="${config:-/www/server/panel/vhost/nginx/api-cn.buggoglobal.com.conf}"
@@ -241,6 +254,31 @@ switch_gateway_proxy() {
   rm -f "$backup"
 }
 
+reload_frontend_proxies() {
+  [ "$SERVICE" = "gateway" ] || [ "$SERVICE" = "gateway-overseas" ] || return 0
+  for frontend in firstleg-prod-admin-web-blue firstleg-prod-admin-web-green \
+    firstleg-prod-client-web-blue firstleg-prod-client-web-green; do
+    if docker inspect "$frontend" >/dev/null 2>&1; then
+      docker exec "$frontend" nginx -t
+      docker exec "$frontend" nginx -s reload
+    fi
+  done
+}
+
+rollback_gateway_proxy() {
+  [ "$SERVICE" = "gateway" ] || [ "$SERVICE" = "gateway-overseas" ] || return 0
+  if [ "$SERVICE" = "gateway" ]; then
+    config="$(env_value GATEWAY_BLUE_GREEN_NGINX_CONFIG)"
+    config="${config:-/www/server/panel/vhost/nginx/api-cn.buggoglobal.com.conf}"
+  else
+    config="$(env_value GATEWAY_OVERSEAS_BLUE_GREEN_NGINX_CONFIG)"
+    config="${config:-/www/server/panel/vhost/nginx/api-global.buggoglobal.com.conf}"
+  fi
+  old_port="${active_port:-$BASE_PORT}"
+  sed -i "s#127\.0\.0\.1:$target_port#127.0.0.1:$old_port#g" "$config"
+  nginx -t && nginx -s reload
+}
+
 bootstrap_legacy_routes
 
 export DEPLOY_ENV IMAGE_REPOSITORY IMAGE_TAG
@@ -276,8 +314,11 @@ if [ "$running_image" != "$IMAGE" ]; then
   exit 1
 fi
 
-connect_routing_alias "$ROUTING_ALIAS" "$candidate"
+if [ "$NACOS_SERVICE" = "$ROUTING_ALIAS" ]; then
-[ "$NACOS_SERVICE" = "$SERVICE" ] || connect_routing_alias "$NACOS_SERVICE" "$candidate"
+  connect_routing_aliases "$candidate" "$ROUTING_ALIAS"
+else
+  connect_routing_aliases "$candidate" "$ROUTING_ALIAS" "$NACOS_SERVICE"
+fi
 
 if [ "$STABILIZATION_SECONDS" -gt 0 ]; then
   echo "候选槽稳定观察 ${STABILIZATION_SECONDS}s"
@@ -290,13 +331,60 @@ if ! switch_gateway_proxy; then
   exit 1
 fi
 
-if [ -n "$active_container" ] && docker inspect "$active_container" >/dev/null 2>&1; then
+worker_required=false
-  echo "优雅停止旧实例: $active_container"
+if needs_worker; then worker_required=true; fi
-  docker stop --time "$STOP_TIMEOUT" "$active_container" >/dev/null
+
-  docker rm "$active_container" >/dev/null
+if [ "$worker_required" = true ] && [ "$active_slot" != legacy ]; then
+  if [ -n "$active_worker" ] && docker inspect "$active_worker" >/dev/null 2>&1; then
+    echo "暂停旧 Worker，避免新旧任务并行: $active_worker"
+    docker stop --time "$STOP_TIMEOUT" "$active_worker" >/dev/null
+  fi
+  if ! start_worker_candidate; then
+    [ -z "$active_worker" ] || docker start "$active_worker" >/dev/null 2>&1 || true
+    rollback_gateway_proxy || true
+    cleanup_candidate
+    exit 1
+  fi
 fi
 
-printf '%s|%s|%s|%s\n' "$target_slot" "$candidate" "$target_port" "$IMAGE" > "$STATE_FILE.tmp"
+old_container_stopped=false
+if [ -n "$active_container" ] && docker inspect "$active_container" >/dev/null 2>&1; then
+  echo "优雅停止旧 Web 实例: $active_container"
+  docker stop --time "$STOP_TIMEOUT" "$active_container" >/dev/null
+  old_container_stopped=true
+fi
+
+if [ "$worker_required" = true ] && [ "$active_slot" = legacy ]; then
+  if ! start_worker_candidate; then
+    [ "$old_container_stopped" = false ] || docker start "$active_container" >/dev/null 2>&1 || true
+    rollback_gateway_proxy || true
+    cleanup_candidate
+    exit 1
+  fi
+fi
+
+if ! reload_frontend_proxies; then
+  [ "$old_container_stopped" = false ] || docker start "$active_container" >/dev/null 2>&1 || true
+  if [ "$worker_required" = true ]; then
+    docker rm -f "$worker_candidate" >/dev/null 2>&1 || true
+    [ -z "$active_worker" ] || docker start "$active_worker" >/dev/null 2>&1 || true
+  fi
+  rollback_gateway_proxy || true
+  cleanup_candidate
+  reload_frontend_proxies || true
+  exit 1
+fi
+
+[ -z "$active_worker" ] || [ "$active_worker" = "$worker_candidate" ] \
+  || docker rm "$active_worker" >/dev/null 2>&1 || true
+[ "$old_container_stopped" = false ] || docker rm "$active_container" >/dev/null
+
+if [ "$worker_required" = true ]; then
+  next_worker="$worker_candidate"
+else
+  next_worker=""
+fi
+printf '%s|%s|%s|%s|%s\n' "$target_slot" "$candidate" "$target_port" "$IMAGE" "$next_worker" > "$STATE_FILE.tmp"
 chmod 640 "$STATE_FILE.tmp"
 mv -f "$STATE_FILE.tmp" "$STATE_FILE"
 echo "蓝绿切换完成: service=$SERVICE active=$target_slot container=$candidate port=$target_port"

deploy/scripts/deploy-built-service.sh

@@ -348,6 +348,16 @@ deploy_local() {
 
 if [ "${TARGET_IS_LOCAL:-false}" = "true" ]; then
   echo "镜像已在目标服务器本机，直接部署: $IMAGE"
+  if [ "$DEPLOY_ENV" = "production" ]; then
+    printf '%s' "$REGISTRY_PASSWORD" | docker login -u "$REGISTRY_USER" --password-stdin "$REGISTRY_HOST"
+    trap local_registry_logout EXIT HUP INT TERM
+    docker pull "$IMAGE"
+    sh deploy/scripts/deploy-blue-green-service.sh \
+      "$DEPLOY_ENV" "$IMAGE_REPOSITORY" "$IMAGE_TAG" "$SERVICE" "$ENV_FILE" "$COMPOSE_FILE" "$PROJECT_NAME"
+    local_registry_logout
+    trap - EXIT HUP INT TERM
+    exit 0
+  fi
   deploy_local
   exit 0
 fi

deploy/scripts/deploy-services.sh

@@ -188,6 +188,19 @@ if [ "$DEPLOY_ENV" = "development" ]; then
     "$IMAGE_REPOSITORY" "$IMAGE_TAG" "$SERVICE_CSV" "$ENV_FILE"
 fi
 
+if [ "$DEPLOY_ENV" = "production" ]; then
+  docker compose \
+    --project-name "$PROJECT_NAME" \
+    --env-file "$ENV_FILE" \
+    -f "$COMPOSE_FILE" \
+    pull $SERVICES
+  for service in $SERVICES; do
+    sh deploy/scripts/deploy-blue-green-service.sh \
+      "$DEPLOY_ENV" "$IMAGE_REPOSITORY" "$IMAGE_TAG" "$service" "$ENV_FILE" "$COMPOSE_FILE" "$PROJECT_NAME"
+  done
+  exit 0
+fi
+
 docker compose \
   --project-name "$PROJECT_NAME" \
   --env-file "$ENV_FILE" \

event-bridge-service/src/main/resources/application.yml

@@ -23,6 +23,8 @@ spring:
       username: ${NACOS_USERNAME:nacos}
       password: ${NACOS_PASSWORD:nacos}
       discovery:
+        enabled: ${NACOS_DISCOVERY_ENABLED:true}
+        register-enabled: ${NACOS_REGISTER_ENABLED:true}
         group: ${NACOS_GROUP:BAIYUE}
         namespace: ${NACOS_NAMESPACE:}
       config:

scripts/blue-green-deployment-contract.test.mjs

@@ -16,17 +16,21 @@ test('all backend business services have blue-green runtime contracts', async ()
   assert.match(script, /target_slot="green"/)
   assert.match(script, /target_port=\$\(\(BASE_PORT \+ 10000\)\)/)
   assert.match(script, /compose run -d --no-deps --name "\$candidate"/)
-  assert.match(script, /wait_for_candidate/)
+  assert.match(script, /wait_for_container "\$candidate"/)
-  assert.match(script, /connect_routing_alias "\$ROUTING_ALIAS" "\$candidate"/)
+  assert.match(script, /BAIYUE_RUNTIME_SCHEDULING_ENABLED=false/)
+  assert.match(script, /NACOS_REGISTER_ENABLED=false/)
+  assert.match(script, /start_worker_candidate/)
+  assert.match(script, /connect_routing_aliases "\$candidate" "\$ROUTING_ALIAS" "\$NACOS_SERVICE"/)
+  assert.match(script, /file,file-service:file/)
   assert.match(script, /docker stop --time "\$STOP_TIMEOUT" "\$active_container"/)
   assert.match(script, /mv -f "\$STATE_FILE\.tmp" "\$STATE_FILE"/)
 })
 
 test('candidate failure keeps the active container and successful cutover stops it last', async () => {
   const script = await read('deploy/scripts/deploy-blue-green-service.sh')
-  const start = script.indexOf('compose run -d')
+  const start = script.lastIndexOf('compose run -d')
-  const healthy = script.indexOf('if ! wait_for_candidate')
+  const healthy = script.indexOf('if ! wait_for_container "$candidate"')
-  const route = script.indexOf('connect_routing_alias "$ROUTING_ALIAS" "$candidate"')
+  const route = script.indexOf('connect_routing_aliases "$candidate"')
   const gateway = script.indexOf('if ! switch_gateway_proxy')
   const stop = script.indexOf('docker stop --time "$STOP_TIMEOUT" "$active_container"')
   const state = script.indexOf('mv -f "$STATE_FILE.tmp" "$STATE_FILE"')
@@ -39,6 +43,8 @@ test('gateway cutover is atomic and restores nginx config on validation failure'
   const script = await read('deploy/scripts/deploy-blue-green-service.sh')
   assert.match(script, /api-cn\.buggoglobal\.com\.conf/)
   assert.match(script, /api-global\.buggoglobal\.com\.conf/)
+  assert.match(script, /尚未使用稳定上游 \$required_alias，拒绝切换 Gateway/)
+  assert.match(script, /reload_frontend_proxies/)
   assert.match(script, /cp -f "\$config" "\$backup"/)
   assert.match(script, /nginx -t \|\| ! nginx -s reload/)
   assert.match(script, /cp -f "\$backup" "\$config"/)
@@ -56,9 +62,27 @@ test('compose templates preserve rolling defaults and expose slot metadata', asy
   }
 })
 
+test('every scheduled worker service can disable Nacos registration', async () => {
+  for (const application of [
+    'iam/src/main/resources/application.yml',
+    'firstleg/src/main/resources/application.yml',
+    'finance/src/main/resources/application.yml',
+    'order/src/main/resources/application.yml',
+    'file/src/main/resources/application.yml',
+    'event-bridge-service/src/main/resources/application.yml',
+    'audit-service/src/main/resources/application.yml',
+    'lastleg-service/src/main/resources/application.yml'
+  ]) {
+    assert.match(await read(application), /register-enabled: \$\{NACOS_REGISTER_ENABLED:true\}/, application)
+  }
+})
+
 test('production single-service pipeline dispatches the blue-green executor', async () => {
   const deploy = await read('deploy/scripts/deploy-built-service.sh')
+  const batch = await read('deploy/scripts/deploy-services.sh')
   assert.match(deploy, /run_scp "deploy\/scripts\/deploy-blue-green-service\.sh"/)
   assert.match(deploy, /if \[ "\$DEPLOY_ENV" = "production" \]/)
   assert.match(deploy, /run_ssh sh "\$REMOTE_DIR\/\$REMOTE_BLUE_GREEN_NAME"/)
+  assert.match(deploy, /TARGET_IS_LOCAL[\s\S]*?deploy-blue-green-service\.sh/)
+  assert.match(batch, /if \[ "\$DEPLOY_ENV" = "production" \][\s\S]*?deploy-blue-green-service\.sh/)
 })

scripts/blue-green-deployment-runtime.test.mjs

@@ -21,12 +21,22 @@ case "$command" in
     exit 0
     ;;
   compose)
+    container_name=''
+    previous=''
     for argument in "$@"; do
+      if [ "$previous" = --name ]; then container_name="$argument"; fi
+      previous="$argument"
       if [ "$argument" = run ]; then
-        : > "$MOCK_CANDIDATE_EXISTS"
+        action=run
-        exit 0
       fi
     done
+    if [ "\${action:-}" = run ]; then
+      case "$container_name" in
+        *-worker) : > "$MOCK_WORKER_EXISTS" ;;
+        *) : > "$MOCK_CANDIDATE_EXISTS" ;;
+      esac
+      exit 0
+    fi
     exit 0
     ;;
   inspect)
@@ -43,6 +53,13 @@ case "$command" in
           *Config.Image*) printf '%s\\n' 'registry.local/group/file:newtag' ;;
         esac
         ;;
+      firstleg-production-file-green-worker)
+        [ -f "$MOCK_WORKER_EXISTS" ] || exit 1
+        case "$*" in
+          *State.Health*) printf '%s\\n' healthy ;;
+          *Config.Image*) printf '%s\\n' 'registry.local/group/file:newtag' ;;
+        esac
+        ;;
       *) exit 1 ;;
     esac
     ;;
@@ -57,6 +74,7 @@ case "$command" in
     case "$target" in
       firstleg-production-file) : > "$MOCK_LEGACY_REMOVED" ;;
       firstleg-production-file-green) rm -f "$MOCK_CANDIDATE_EXISTS" ;;
+      firstleg-production-file-green-worker) rm -f "$MOCK_WORKER_EXISTS" ;;
     esac
     ;;
   logs) echo 'candidate logs' ;;
@@ -79,13 +97,13 @@ async function runBlueGreen(health) {
   await writeFile(sleep, '#!/usr/bin/env sh\nexit 0\n', 'utf8')
   await chmod(docker, 0o755)
   await chmod(sleep, 0o755)
-  await writeFile(envFile, 'PLACEHOLDER=true\n', 'utf8')
+  await writeFile(envFile, 'FILE_EXPORT_CONSUMER_ENABLED=true\n', 'utf8')
   await writeFile(compose, 'services: {}\n', 'utf8')
   await writeFile(log, '', 'utf8')
 
   const paths = {
     candidate: join(temp, 'candidate'), stopped: join(temp, 'legacy-stopped'),
-    removed: join(temp, 'legacy-removed')
+    removed: join(temp, 'legacy-removed'), worker: join(temp, 'worker')
   }
   const result = spawnSync(shell, [
     script.replaceAll('\\', '/'), 'production', 'registry.local/group', 'newtag', 'file',
@@ -100,6 +118,7 @@ async function runBlueGreen(health) {
       DEPLOY_HEALTH_POLL_SECONDS: '1',
       MOCK_LOG: log,
       MOCK_CANDIDATE_EXISTS: paths.candidate,
+      MOCK_WORKER_EXISTS: paths.worker,
       MOCK_LEGACY_STOPPED: paths.stopped,
       MOCK_LEGACY_REMOVED: paths.removed,
       MOCK_CANDIDATE_HEALTH: health
@@ -111,7 +130,8 @@ async function runBlueGreen(health) {
   const state = existsSync(stateFile) ? await readFile(stateFile, 'utf8') : ''
   const evidence = {
     ...result, output, dockerLog, state,
-    stopped: existsSync(paths.stopped), removed: existsSync(paths.removed)
+    stopped: existsSync(paths.stopped), removed: existsSync(paths.removed),
+    worker: existsSync(paths.worker)
   }
   await rm(temp, { recursive: true, force: true })
   return evidence
@@ -123,10 +143,13 @@ test('healthy candidate becomes active only after health and old instance drains
   assert.equal(result.status, 0, result.output)
   assert.equal(result.stopped, true)
   assert.equal(result.removed, true)
-  assert.match(result.state, /^green\|firstleg-production-file-green\|19206\|registry\.local\/group\/file:newtag/)
+  assert.equal(result.worker, true)
+  assert.match(result.state, /^green\|firstleg-production-file-green\|19206\|registry\.local\/group\/file:newtag\|firstleg-production-file-green-worker/)
   const run = result.dockerLog.indexOf('compose --project-name firstleg-production')
   const stop = result.dockerLog.indexOf('stop --time 45 firstleg-production-file')
   assert.ok(run >= 0 && run < stop)
+  assert.match(result.dockerLog,
+    /network connect --alias file --alias file-service firstleg-production-routing firstleg-production-file-green/)
 })
 
 test('unhealthy candidate is removed while old instance and state remain untouched', async (context) => {
@@ -135,6 +158,7 @@ test('unhealthy candidate is removed while old instance and state remain untouch
   assert.equal(result.status, 1, result.output)
   assert.equal(result.stopped, false)
   assert.equal(result.removed, false)
+  assert.equal(result.worker, false)
   assert.equal(result.state, '')
   assert.doesNotMatch(result.dockerLog, /stop --time .*firstleg-production-file/)
 })