API网关将响应从403随机切换到404

我已设置API网关，以使用Lambda为S3存储桶中的某些文件提供服务。当我尝试请求不存在的文件时，API网关有时会以403 Forbidden响应（大多数情况下甚至不会触发Lambda函数），有时会以404 Not Found错误响应（我想在这种情况下触发404） ）。

我的Lambda函数非常简单：

exports.handler = async event => {
  try {
    const Bucket = 'testing-bucket';
    const Key = `${event.documentType}/${event.requestParams.document}`;
    const file = await s3.getObject({ Bucket, Key }).promise();

    return {
      body: file.Body.toString('base64'),
      headers: {
        'Content-Disposition': `attachment; filename=test.jpg`,
        'Content-Length': file.ContentLength,
        'Content-Type': file.ContentType,
      },
      statusCode: 200,
      isBase64Encoded: true,
    };
  } catch (e) {
    return {
      body: JSON.stringify({ message: `[${e.code}] ${e.message}` }),
      statusCode: e.statusCode,
    };
  }
};

以这种方式配置附加到Lambda函数的IAM角色：

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Sid": "VisualEditor0",
            "Effect": "Allow",
            "Action": [
                "s3:PutObject",
                "s3:GetObject",
                "s3:ListBucket"
            ],
            "Resource": [
                "arn:aws:s3:::testing-bucket",
                "arn:aws:s3:::testing-bucket/*"
            ]
        }
    ]
}

API网关中的缓存已完全禁用，我一直在尝试测试的命令是：

curl -X GET -H 'Authorization: 123xyz' -H 'Accept: image/jpeg' -H 'Cache-Control: no-cache' -I .xxx

回复为：

HTTP/2 403
content-type: application/json
content-length: 60
date: Mon, 07 Oct 2019 10:32:30 GMT
x-amzn-requestid: ae870104-9045-4c23-9794-226992bad591
x-amzn-errortype: AccessDeniedException
x-amz-apigw-id: BMAZwGSyoAMFftw=
x-cache: Error from cloudfront
via: 1.1 ccf34ecc11e5579d8083b17d9d39a622.cloudfront (CloudFront)
x-amz-cf-pop: LHR62-C2
x-amz-cf-id: zgtgfJX9TQLcI8F2RLWdgTz-RN_1j7MXblQ1498ucoeFY3dhjitOdg==

和

HTTP/2 404
content-type: application/json
content-length: 59
date: Mon, 07 Oct 2019 10:32:31 GMT
x-amzn-requestid: 2de49681-4f21-4cd1-989c-9b36327badb1
x-amz-apigw-id: BMAZ5E52IAMFwEg=
x-amzn-trace-id: Root=1-5d9b143f-aadf0a24a5f60f4c939b77c0;Sampled=0
x-cache: Error from cloudfront
via: 1.1 be00537a2361673ea48963d6e04d04a1.cloudfront (CloudFront)
x-amz-cf-pop: LHR62-C2
x-amz-cf-id: 9VI26GH3-ZuJSQrEt5Fc7EjuMt8IV0TPzPwna8dvvr6UtsgiqwwIkw==

如何使API网关以一致的方式响应？

更新：

[观察API网关日志并尝试连续两次对现有和不存在的文件进行相同的curl垃圾邮件之后，这是不存在的文件的输出（时间戳是完整的：]]]

# curl -X GET -H 'Authorization: 123xyz' -H 'Accept: image/jpeg' -H 'cache-control: private, no-cache, no-store, max-age=1, s-maxage=1'  /foo/nobar

{
    "requestId": "d19602e8-3a32-4445-b9e6-99f05a59fac4",
    "ip": "redacted",
    "caller": "-",
    "user": "-",
    "requestTime": "08/Oct/2019:00:05:03 +0000",
    "httpMethod": "GET",
    "resourcePath": "/foo/{bar}",
    "status": "404",
    "protocol": "HTTP/1.1",
    "responseLength": "59"
}

# and

{
    "requestId": "b33bf6c7-55db-4e1f-b4e4-b1e826139556",
    "ip": "redacted",
    "caller": "-",
    "user": "-",
    "requestTime": "08/Oct/2019:00:05:05 +0000",
    "httpMethod": "GET",
    "resourcePath": "/foo/{bar}",
    "status": "403",
    "protocol": "HTTP/1.1",
    "responseLength": "60"
}
并且对于现有文件：

# curl -X GET -H 'Authorization: 123xyz' -H 'Accept: image/jpeg' -H 'cache-control: private, no-cache, no-store, max-age=1, s-maxage=1'  /foo/bar

{
    "requestId": "122ef31e-c587-470c-a0b5-51c6d9838fe4",
    "ip": "redacted",
    "caller": "-",
    "user": "-",
    "requestTime": "07/Oct/2019:23:58:35 +0000",
    "httpMethod": "GET",
    "resourcePath": "/foo/{bar}",
    "status": "403",
    "protocol": "HTTP/1.1",
    "responseLength": "60"
}

# and then later

{
    "requestId": "c8ad1b40-006f-4d03-9d10-c6d91e366380",
    "ip": "redacted",
    "caller": "-",
    "user": "-",
    "requestTime": "07/Oct/2019:23:59:58 +0000",
    "httpMethod": "GET",
    "resourcePath": "/foo/{bar}",
    "status": "200",
    "protocol": "HTTP/1.1",
    "responseLength": "80280"
}

我已设置API网关，以使用Lambda为S3存储桶中的某些文件提供服务。当我尝试请求不存在的文件时，API网关有时会以“ 403禁止”响应（大多数情况下，它不会...

您可以自行管理回复。您可以在等待后检查文件是否不存在并响应404。类似这样的内容：（未测试代码）