Throttling errors - Amazon EMR
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).

Throttling errors

The errors "Throttled from Amazon EC2 while launching cluster" and "Failed to provision instances due to throttling from Amazon EC2" occur when Amazon EMR cannot complete a request because another service has throttled the activity. Amazon EC2 is the most common source of throttling errors, but other services may be the cause of throttling errors. Amazon service limits apply on a per-Region basis to improve performance, and a throttling error indicates that you have exceeded the service limit for your account in that Region.

Possible causes

The most common source of Amazon EC2 throttling errors is a large number of cluster instances launching so that your service limit for EC2 instances is exceeded. Cluster instances may launch for the following reasons:

It is also possible that the frequency or type of API request being made to Amazon EC2 causes throttling errors. For more information about how Amazon EC2 throttles API requests, see Query API request rate in the Amazon EC2 API Reference.


Consider the following solutions:

  • Follow the instructions in Amazon service quotas in the Amazon Web Services General Reference to request a service limit increase. For some APIs, setting up a CloudWatch event might be a better option than increasing limits. For more details, see When to set up EMR events in CloudWatch.

  • If you have clusters that launch on the same schedule—for example, at the top of the hour—consider staggering start times.

  • If you have clusters that are sized for peak demand, and you periodically have instance capacity, consider specifying automatic scaling to add and remove instances on-demand. In this way, instances are used more efficiently, and depending on the demand profile, fewer instances may be requested at a given time across an account. For more information, see Using automatic scaling with a custom policy for instance groups.