在 Cloudfront 后面为 S3 提供 2 个来源

Question

I've a CF distribution with 2 origins:

• S3
• ALB

This distribution has an alternate CNAME.

Depending on the path, I want some files being served by S3, other by the ALB.

I've 3 behaviours with the following paths:

• assets/* with S3 as origin
• uploads/* with the alb as origin
• default with the same uploads/* 's origin but with a different cache policy

In the S3 bucket there are 2 folders: svg and v2

when I try
curl -I https://myalternate.domain.name/assets/svg/mysvg.svg or
curl -I https://xxxxxx.cloudfront.net/assets/svg/mysvg.svg
I've 404.

If I get them from web I can see

<Error>
  <Code>NoSuchBucket</Code>
  <Message>The specified bucket does not exist</Message>
  <BucketName>myalternate.domain.name or cloudfront endpoint</BucketName>
  <RequestId>Q7Z1NAP8HV0PCEM4</RequestId>
</Error>

For the other paths everything is alright.

What did I miss?

Answer 1

What the CloudFront does when it receives the https://myalternate.domain.name/assets/svg/mysvg.svg request is that it will fetch the object stored in S3 using the key assets/svg/mysvg.svg . That's why it returns 404 since the actual key of that object is svg/mysvg.svg . The solution is to either add assets/ prefix to every object in S3 or use a Lambda@Edge function to rewrite the origin request.