fix(lib-storage): call AbortMultipartUpload when failing to CompleteMultipartUpload

[kuhe]

Issue

related to https://github.com/aws/aws-sdk-js-v3/pull/4414

• [x] include fix for https://github.com/aws/aws-sdk-js-v3/issues/5964 as well

Description

This PR modifies the behavior of @aws-sdk/lib-storage's Upload class' leavePartsOnError: boolean to do what it sounds like it does instead of something meaningless.

Previous to this PR:

• when true, any error encountered during MPU would be thrown
• when false, any error encountered during MPU would be ignored
• default was false

As of this PR:

• when true, AbortMultipartUpload is not called in the event of a failed MPU
• when false, AbortMultipartUpload is called in the event of a failed MPU to mark the parts for cleanup as recommended by S3
• default is still false

Testing

new e2e test

Additional context

Portion of docs from AWS SDK JSv2:

 * ## Handling Multipart Cleanup
 *
 * By default, this class will automatically clean up any multipart uploads
 * when an individual part upload fails. This behavior can be disabled in order
 * to manually handle failures by setting the `leavePartsOnError` configuration
 * option to `true` when initializing the upload object.

behavior (JSv2 source code):

    if (self.service.config.params.UploadId && !self.leavePartsOnError) {
      self.service.abortMultipartUpload().send();
    } else if (self.leavePartsOnError) {
      self.isDoneChunking = false;
    }

This behavior appears to have been lost in https://github.com/aws/aws-sdk-js-v3/pull/2039

Checklist

• [x] If you wrote E2E tests, are they resilient to concurrent I/O?

[github-actions[bot]]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs and link to relevant comments in this thread.