fix(deploy): 修复生产部署脚本导致web/worker启动失败的502问题

- worker: 添加workdir /app/apps/worker + entrypoint-worker.sh入口，正确启动双celery队列 - worker: 添加PYTHONPATH、双network-alias(xiaoxia-worker/xiaoxia-api)、legacy-assets卷 - worker: 健康检查改用celery inspect ping（容器无ps命令） - worker: 内存限制从2g调整为3g，日志上限提升至200m - web: 添加APP_ENV/API_BASE_URL环境变量和network-alias xiaoxia-web - api: 添加network-alias xiaoxia-api，修正GENERATED_FILES_URL_PREFIX为完整URL - 修正PUBLIC_API_BASE_URL错误域名(production-api→saas-api) - 回滚段同步修复，保证回滚也能正确启动 - 修正结尾端口打印错误(8000/3001→8001/3002)
2026-09-14 11:20:22 +08:00
parent 96b3d6ae12
commit c3fdbd9e65
1 changed files with 52 additions and 38 deletions
@@ -170,16 +170,17 @@ rollback() {
      --name xiaoxia-api-production \
      --env-file "$ENV_FILE" \
      --network xiaoxia-net-production \
+      --network-alias xiaoxia-api \
      -p 127.0.0.1:8001:8000 \
      -e APP_ENV=production \
      -e APP_VERSION="$(echo $PREV_API_IMAGE | grep -oE '[^:]+$')" \
      -e GENERATED_FILES_DIR=/app/generated \
-      -e GENERATED_FILES_URL_PREFIX=/generated-files \
-      -e PUBLIC_API_BASE_URL=https://production-api.xiaoxiajianji.com \
+      -e GENERATED_FILES_URL_PREFIX=https://saas-api.xiaoxiajianji.com/generated-files \
+      -e PUBLIC_API_BASE_URL=https://saas-api.xiaoxiajianji.com \
      -v "$GENERATED_DIR:/app/generated" \
      --restart unless-stopped \
-  --cpus 2 \
-  --memory 2g \
+      --cpus 2 \
+      --memory 2g \
      --health-cmd "python -c \"import urllib.request; urllib.request.urlopen('http://localhost:8000/health', timeout=5)\"" \
      --health-interval 30s \
      --health-timeout 10s \
@@ -196,26 +197,29 @@ rollback() {
    echo "Rolling back Worker to: $PREV_WORKER_IMAGE"
    docker run -d \
      --name xiaoxia-worker-production \
-      --env-file "$ENV_FILE" \
      --network xiaoxia-net-production \
+      --network-alias xiaoxia-worker \
+      --network-alias xiaoxia-api \
+      --env-file "$ENV_FILE" \
      -e APP_ENV=production \
      -e APP_VERSION="$(echo $PREV_WORKER_IMAGE | grep -oE '[^:]+$')" \
-      -e WORKER_CONCURRENCY=1 \
-      -e WORKER_MAX_TASKS_PER_CHILD=100 \
      -e GENERATED_FILES_DIR=/app/generated \
-      -e GENERATED_FILES_URL_PREFIX=/generated-files \
-      -e PUBLIC_API_BASE_URL=https://production-api.xiaoxiajianji.com \
+      -e GENERATED_FILES_URL_PREFIX=https://saas-api.xiaoxiajianji.com/generated-files \
+      -e PYTHONPATH=/app:/app/apps/api:/app/packages \
      -v "$GENERATED_DIR:/app/generated" \
+      -v "$LEGACY_ASSETS_DIR:/app/legacy-assets" \
+      -w /app/apps/worker \
      --restart unless-stopped \
-  --cpus 2 \
-  --memory 2g \
-      --health-cmd "sh -c \"for pid in /proc/[0-9]*/cmdline; do if grep -ql celery \"$pid\" 2>/dev/null; then exit 0; fi; done; exit 1\"" \
+      --cpus 2 \
+      --memory 3g \
+      --health-cmd "sh -c 'PYTHONPATH=/app:/app/apps/api:/app/packages celery -A worker_app.celery_app inspect ping -t 5 2>&1 | grep -q pong'" \
      --health-interval 30s \
-      --health-timeout 10s \
+      --health-timeout 15s \
      --health-retries 3 \
-      --health-start-period 30s \
-      $LOG_OPTS \
-      "$PREV_WORKER_IMAGE"
+      --health-start-period 60s \
+      --log-driver json-file --log-opt max-size=200m --log-opt max-file=5 \
+      "$PREV_WORKER_IMAGE" \
+      /usr/local/bin/entrypoint-worker.sh
  else
    echo "No previous Worker image to roll back to"
  fi
@@ -230,12 +234,15 @@ rollback() {
    docker run -d \
      --name xiaoxia-web-production \
      --network xiaoxia-net-production \
+      --network-alias xiaoxia-web \
      -p 127.0.0.1:3002:80 \
-      --restart unless-stopped \
-  --cpus 0.5 \
-  --memory 512m \
-      $LEGACY_VOLUME \
+      -e APP_ENV=production \
+      -e API_BASE_URL=https://saas-api.xiaoxiajianji.com \
      -v "$NGINX_CONF_FILE:/etc/nginx/conf.d/default.conf:ro" \
+      $LEGACY_VOLUME \
+      --restart unless-stopped \
+      --cpus 1 \
+      --memory 512m \
      --health-cmd "wget --spider -q http://127.0.0.1:80" \
      --health-interval 30s \
      --health-timeout 5s \
@@ -373,12 +380,13 @@ docker run -d \
  --name xiaoxia-api-production \
  --env-file "$ENV_FILE" \
  --network xiaoxia-net-production \
+  --network-alias xiaoxia-api \
  -p 127.0.0.1:8001:8000 \
  -e APP_ENV=production \
  -e APP_VERSION="$IMAGE_TAG" \
  -e GENERATED_FILES_DIR=/app/generated \
-  -e GENERATED_FILES_URL_PREFIX=/generated-files \
-  -e PUBLIC_API_BASE_URL=https://production-api.xiaoxiajianji.com \
+  -e GENERATED_FILES_URL_PREFIX=https://saas-api.xiaoxiajianji.com/generated-files \
+  -e PUBLIC_API_BASE_URL=https://saas-api.xiaoxiajianji.com \
  -v "$GENERATED_DIR:/app/generated" \
  --restart unless-stopped \
  --cpus 2 \
@@ -395,26 +403,29 @@ docker run -d \
 echo "Starting Worker container..."
 docker run -d \
  --name xiaoxia-worker-production \
-  --env-file "$ENV_FILE" \
  --network xiaoxia-net-production \
+  --network-alias xiaoxia-worker \
+  --network-alias xiaoxia-api \
+  --env-file "$ENV_FILE" \
  -e APP_ENV=production \
  -e APP_VERSION="$IMAGE_TAG" \
-  -e WORKER_CONCURRENCY=1 \
-  -e WORKER_MAX_TASKS_PER_CHILD=100 \
  -e GENERATED_FILES_DIR=/app/generated \
-  -e GENERATED_FILES_URL_PREFIX=/generated-files \
-  -e PUBLIC_API_BASE_URL=https://production-api.xiaoxiajianji.com \
+  -e GENERATED_FILES_URL_PREFIX=https://saas-api.xiaoxiajianji.com/generated-files \
+  -e PYTHONPATH=/app:/app/apps/api:/app/packages \
  -v "$GENERATED_DIR:/app/generated" \
+  -v "$LEGACY_ASSETS_DIR:/app/legacy-assets" \
+  -w /app/apps/worker \
  --restart unless-stopped \
  --cpus 2 \
-  --memory 2g \
-  --health-cmd "sh -c \"for pid in /proc/[0-9]*/cmdline; do if grep -ql celery \"$pid\" 2>/dev/null; then exit 0; fi; done; exit 1\"" \
+  --memory 3g \
+  --health-cmd "sh -c 'PYTHONPATH=/app:/app/apps/api:/app/packages celery -A worker_app.celery_app inspect ping -t 5 2>&1 | grep -q pong'" \
  --health-interval 30s \
-  --health-timeout 10s \
+  --health-timeout 15s \
  --health-retries 3 \
-  --health-start-period 30s \
-  $LOG_OPTS \
-  "$REGISTRY_WORKER" || rollback
+  --health-start-period 60s \
+  --log-driver json-file --log-opt max-size=200m --log-opt max-file=5 \
+  "$REGISTRY_WORKER" \
+  /usr/local/bin/entrypoint-worker.sh || rollback

 # ---- 启动 Web ----
 LEGACY_VOLUME=""
@@ -429,12 +440,15 @@ echo "Starting Web container..."
 docker run -d \
  --name xiaoxia-web-production \
  --network xiaoxia-net-production \
+  --network-alias xiaoxia-web \
  -p 127.0.0.1:3002:80 \
-  --restart unless-stopped \
-  --cpus 0.5 \
-  --memory 512m \
+  -e APP_ENV=production \
+  -e API_BASE_URL=https://saas-api.xiaoxiajianji.com \
  -v "$NGINX_CONF_FILE:/etc/nginx/conf.d/default.conf:ro" \
  $LEGACY_VOLUME \
+  --restart unless-stopped \
+  --cpus 1 \
+  --memory 512m \
  --health-cmd "wget --spider -q http://127.0.0.1:80" \
  --health-interval 30s \
  --health-timeout 5s \
@@ -487,7 +501,7 @@ docker builder prune -af --filter "until=168h" 2>/dev/null || true

 echo ""
 echo "=== Production deployment complete ==="
-echo "API:      http://127.0.0.1:8000"
-echo "Web:      http://127.0.0.1:3001"
+echo "API:      http://127.0.0.1:8001"
+echo "Web:      http://127.0.0.1:3002"
 echo "Version:  $IMAGE_TAG"
 docker ps --format "table {{.Names}}\t{{.Status}}\t{{.Image}}" | grep production