如何用TypeScript CDK部署含外部库的Python Lambda及多Lambda最佳实践

处理外部依赖（以requests库为例） #

• 先在Python Lambda的代码目录下创建requirements.txt，明确指定依赖版本，比如：

  requests==2.31.0
  

• 有两种打包方式可选：
  1. 直接通过PythonFunction自动打包：利用CDK的PythonFunction（来自aws-cdk-lib/aws-lambda-python-alpha模块）的bundling配置，让CDK自动把依赖和代码打包在一起，默认会用Docker模拟Lambda运行环境，避免本地依赖不兼容问题：

     import { PythonFunction } from 'aws-cdk-lib/aws-lambda-python-alpha';
     import { Stack, StackProps } from 'aws-cdk-lib';
     import { Construct } from 'constructs';
     import * as lambda from 'aws-cdk-lib/aws-lambda';
     
     export class MyStack extends Stack {
       constructor(scope: Construct, id: string, props?: StackProps) {
         super(scope, id, props);
     
         new PythonFunction(this, 'RequestHandlerLambda', {
           entry: './lambda-functions/request-handler', // Python代码所在目录
           runtime: lambda.Runtime.PYTHON_3_11,
           bundling: {
             command: [
               'bash', '-c',
               'pip install -r requirements.txt -t /asset-output && cp -au . /asset-output'
             ],
           },
         });
       }
     }
     

  2. 构建共享依赖层：如果多个Lambda都需要用requests这类库，把依赖打包成Layer，复用给所有需要的Lambda，减少重复打包：

     import { PythonLayerVersion } from 'aws-cdk-lib/aws-lambda-python-alpha';
     import * as lambda from 'aws-cdk-lib/aws-lambda';
     
     // 先创建依赖层
     const requestsLayer = new PythonLayerVersion(this, 'RequestsCommonLayer', {
       entry: './layers/common-requests', // 该目录下放置requirements.txt
       compatibleRuntimes: [lambda.Runtime.PYTHON_3_11],
     });
     
     // 给Lambda挂载层
     new PythonFunction(this, 'AnotherLambda', {
       entry: './lambda-functions/another-handler',
       runtime: lambda.Runtime.PYTHON_3_11,
       layers: [requestsLayer],
     });
     

多Lambda的推荐实践与目录结构 #

目录结构示例 #

my-cdk-project/
├── bin/
│   └── my-cdk-app.ts       # CDK应用入口，定义要部署的栈
├── lib/
│   ├── stacks/
│   │   ├── api-service-stack.ts    # 存放API接口类Lambda的栈
│   │   └── data-processing-stack.ts # 存放后台数据处理类Lambda的栈
│   └── constructs/
│       └── base-python-lambda.ts # 封装通用配置的自定义Construct，复用Lambda基础设置
├── lambda-functions/
│   ├── api-handlers/
│   │   ├── get-user/
│   │   │   ├── index.py
│   │   │   └── requirements.txt
│   │   └── create-user/
│   │       ├── index.py
│   │       └── requirements.txt
│   └── processing-jobs/
│       ├── data-transform/
│       │   ├── index.py
│       │   └── requirements.txt
│       └── notification-sender/
│           ├── index.py
│           └── requirements.txt
├── layers/
│   └── common-deps/
│       └── requirements.txt # 存放requests、boto3等多Lambda共用的依赖
└── cdk.json

栈的拆分原则 #

• 按业务域拆分：比如把API相关的Lambda集中放在ApiServiceStack，数据处理类的放在DataProcessingStack，便于权限管理和资源隔离
• 按资源关联性拆分：如果多个Lambda需要共享同一个DynamoDB表、S3桶，建议放在同一栈内，减少跨栈引用的复杂度
• 按部署频率拆分：变更频繁的Lambda（比如API接口）单独放一个栈，稳定的后台任务Lambda放另一个栈，避免不必要的全栈部署

复用Lambda配置 #

• 封装自定义Construct：把Python Lambda的通用配置（比如运行时、默认日志级别、超时时间）封装成一个Construct，避免重复写相同代码：

  import { Construct } from 'constructs';
  import { PythonFunction } from 'aws-cdk-lib/aws-lambda-python-alpha';
  import { Runtime } from 'aws-cdk-lib/aws-lambda';
  
  export class BasePythonLambda extends Construct {
    constructor(scope: Construct, id: string, props: { entry: string, layers?: any[] }) {
      super(scope, id);
  
      new PythonFunction(this, 'LambdaFunction', {
        entry: props.entry,
        runtime: Runtime.PYTHON_3_11,
        layers: props.layers,
        environment: {
          LOG_LEVEL: 'INFO',
        },
        timeout: cdk.Duration.seconds(30),
        memorySize: 256,
      });
    }
  }
  

• 共用依赖层：把多个Lambda都需要的库打包成共享层，减少每个Lambda的包体积，提升部署效率

内容的提问来源于stack exchange，提问作者Some-one