Best Practices for Amazon Development with the Amazon SDK for Java - Amazon SDK for Java 1.x
S3
Services or capabilities described in Amazon Web Services documentation might vary by Region. To see the differences applicable to the China Regions, see Getting Started with Amazon Web Services in China (PDF).

We announced the upcoming end-of-support for Amazon SDK for Java (v1). We recommend that you migrate to Amazon SDK for Java v2. For dates, additional details, and information on how to migrate, please refer to the linked announcement.

Best Practices for Amazon Development with the Amazon SDK for Java

The following best practices can help you avoid issues or trouble as you develop Amazon applications with the Amazon SDK for Java. We’ve organized best practices by service.

S3

Avoid ResetExceptions

When you upload objects to Amazon S3 by using streams (either through an AmazonS3 client or TransferManager), you might encounter network connectivity or timeout issues. By default, the Amazon SDK for Java attempts to retry failed transfers by marking the input stream before the start of a transfer and then resetting it before retrying.

If the stream doesn’t support mark and reset, the SDK throws a ResetException when there are transient failures and retries are enabled.

Best Practice

We recommend that you use streams that support mark and reset operations.

The most reliable way to avoid a ResetException is to provide data by using a File or FileInputStream, which the Amazon SDK for Java can handle without being constrained by mark and reset limits.

If the stream isn’t a FileInputStream but does support mark and reset, you can set the mark limit by using the setReadLimit method of RequestClientOptions. Its default value is 128 KB. Setting the read limit value to one byte greater than the size of stream will reliably avoid a ResetException.

For example, if the maximum expected size of a stream is 100,000 bytes, set the read limit to 100,001 (100,000 + 1) bytes. The mark and reset will always work for 100,000 bytes or less. Be aware that this might cause some streams to buffer that number of bytes into memory.