Amazon S3 path-based URLs deprecation

Last year Amazon AWS announced plans to deprecate the path-style access to the S3 buckets. In that model, Amazon S3 URLs follow the format where bucket name is specified in the path part of the URL, as shown below:

<i>https://s3.</i><i>Region</i><i></i><i>bucket-name</i><i>/</i><i>key name</i>

For some time now, Amazon AWS has provided an alternate way to access S3 resources using virtual hosted-style URLs. In this model, the bucket name is part of the hostname:

<i>https://</i><i>bucket-name</i><i>.s3.</i><i>Region</i><i></i><i>key name</i><i> </i>

You can learn more about path-style and virtual hosted-style URLs from S3 documentation.

Buckets created after September 30, 2020, will support only virtual hosted-style requests. Path-style requests will continue to be supported for buckets created on or before this date.

You can read the full story and reasoning behind the change in the great blog post written by AWS Chief Evangelist Jeff Bar.

Starting September 1st, 2020, the Filestack Upload API will begin using virtual hosted style URLs in all our communication from File Picker to S3.

We have been working hard to prepare for the upcoming changes and to make sure there is no impact to our customers and their users. However, if you rely on the old path-style model of S3 URLs in any part of your environment, you will have to update to the new schema by that date. This is especially important if you:

  • have users behind corporate firewalls where specific domains must be whitelisted
  • use Content Security Policies (CSP) to control what CORS requests can originate from your applications

As an exception, we will still use path-style URLs for buckets with dots in their names to ensure compatibility and TLS encryption.

Please do not hesitate to reach out to our support team. We are here to help with any questions or concerns.

Contact Support

Read More →