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).

Retry Amazon S3 requests with EMRFS

This topic provides information about the retry strategies that you can use when making requests to Amazon S3 with EMRFS. When your request rate increases, S3 tries to scale to support the new rate. During this process S3 can throttle requests and return a 503 Slow Down error. To improve the success rate of your S3 requests, you can adjust your retry strategy by configuring properties in your emrfs-site configuration.

You can adjust your retry strategy in the following ways.

Use the default exponential backoff strategy

By default, EMRFS uses an exponential backoff strategy to retry Amazon S3 requests. The default EMRFS retry limit is 15. To avoid an S3 503 Slow Down error, you can increase the retry limit when you create a new cluster, on a running cluster, or at application runtime.

To increase the retry limit, you must change the value for fs.s3.maxRetries in your emrfs-site configuration. The following example configuration sets fs.s3.maxRetries to a custom value of 30.

[
    {
      "Classification": "emrfs-site",
      "Properties": {
        "fs.s3.maxRetries": "30"
      }
    }
]

For more information about working with configuration objects, see Configure applications.

Use the AIMD retry strategy

With Amazon EMR release 6.4.0 and later, EMRFS supports an alternative retry strategy based on an additive-increase/multiplicative-decrease (AIMD) model. The AIMD retry strategy is especially useful when you work with large Amazon EMR clusters.

AIMD calculates a custom request rate using data about recent successful requests. This strategy decreases the number of throttled requests and the total attempts required per request.

To enable the AIMD retry strategy, you must set the fs.s3.aimd.enabled property to true in your emrfs-site configuration as in the following example.

[
    {
      "Classification": "emrfs-site",
      "Properties": {
        "fs.s3.aimd.enabled": "true"
      }
    }
]

For more information about working with configuration objects, see Configure applications.

Advanced AIMD retry settings

You can configure the properties listed in the following table to refine retry behavior when you use the AIMD retry strategy. For most use cases, we recommend that you use the default values.