Disabling S3 global endpoint usage via configuration

[roded]

Hello,
I have the following S3Client setup code:

StaticCredentialsProvider credentialsProvider =
    StaticCredentialsProvider.create(AwsBasicCredentials.create("...", "..."));
S3Client s3Client = S3Client.builder()
    .region(Region.US_EAST_1)
    .credentialsProvider(credentialsProvider)
    .build();

s3Client.listObjectsV2(ListObjectsV2Request.builder().bucket("my-bucket").build());

As is, the SDK attempts to access host: my-bucket.s3.amazonaws.com.

Is it possible to configure the JVM via environment variables or system properties to configure the SDK to access the bucket host including the region name? e.g.: my-bucket.s3.us-east-1.amazonaws.com.

Thanks

[debora-ito]

The Java SDK will use the regional s3 endpoint for all regions except us-east-1. That's because requests made with the legacy global endpoint will be routed to the us-east-1 region automatically by S3. For more info on the legacy global endpoint check the S3 docs: https://docs.aws.amazon.com/AmazonS3/latest/userguide/VirtualHosting.html#VirtualHostingBackwardsCompatibility

[roded]

Thanks @debora-ito .
And there's no way to control such behavior via system properties or environment variables?
I.e., making the Java SDK use the regional endpoint for us-east-1 instead of the legacy global endpoint.
Some customers have FW rules blocking access to the global endpoint so only the regional endpoints work.
Thanks

[debora-ito]

No, there's no system property or environment variable to configure this.

One workaround is to use the regional us-east-1 endpoint in the client's EndpointOverride, but you'd need to make the change in code, there's no environment variable or system property for this either.

[roded]

Understood. Yes, that was my chosen solution.
Please consider allowing configuring such behavior via env vars and/or a system property.
Thanks @debora-ito .

[github-actions[bot]]

Hello! Reopening this discussion to make it searchable.