Storage.get( { download: true } ) does not get latest version of S3 bucket file after modifying

Unsure if this is a bug, but encountered some weird caching behaviour with Storage.get() when retrieving file data contents, so this should be clarified. I have an Amplify project that gets data from a text file stored in an S3 bucket made with initializing Amplify storage. The issue is when recent updates are made on a specific file, like updating the file directly on the management console or calling Storage.put(), Calling Storage.get() with { download: true } to get the file contents will still retrieve the unmodified version of the file, and not the latest version; this issue persists for some short period of time until get() eventually retrieves the latest. I see that the changes are reflected in the metadata of the file when I call Storage.list(), so I am not sure why Storage.get does not get the changed file immediately. I doubt if this is expected behaviour, why does this happen? Any way to resolve this issue?

Project was initialized with amplify init and S3 storage resources were made by calling amplify add storage.

A short example of how I use Storage.get():

Storage.get('file-name', { download: true }) .then(data => { data.Body.text().then(string => { // handle the String data return String }); }) .catch(e => { // handle error });

[wei]

Same as #7095

By default, the HTTP Headers returned by S3 looks something like follows:

HTTP/1.1 200 OK
Content-Type: application/json
Access-Control-Allow-Origin: *
Last-Modified: Wed, 18 Nov 2020 16:53:16 GMT
Access-Control-Allow-Methods: GET, HEAD, PUT, POST, DELETE
Access-Control-Expose-Headers: x-amz-server-side-encryption, x-amz-request-id, x-amz-id-2, ETag
Date: Wed, 18 Nov 2020 17:49:49 GMT
Content-Length: 14
Accept-Ranges: bytes
ETag: "XXXXXXXX"
Vary: Origin, Access-Control-Request-Headers, Access-Control-Request-Method
x-amz-id-2: XXXXXXXXxX
Server: AmazonS3
x-amz-request-id: XXXXXXXXXX

Note there is no Cache-Control or Expires headers. When those headers are missing AND Last-Modified header is present, browsers calculates what’s called Heuristic freshness and sets its own ttl in cache. e.g. Firefox docs

This results in subsequent requests getting cached by default as seen:

As Harry mentioned in https://github.com/aws-amplify/amplify-js/issues/7095#issuecomment-728355794 Doing something like

Storage.get(filename, { download: true, cacheControl: 'no-cache' })

will get S3 to return the appropriate Cache-Control headers which eliminates this issue.

[Jaftem]

Similar issue (that’s also unresolved) here: #6413