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).



When a cluster terminates with a BOOTSTRAP_FAILURE_PRIMARY_WITH_NON_ZERO_CODE error, a bootstrap action has failed in the primary instance. For more information about bootstrap actions, see Create bootstrap actions to install additional software.


To resolve this error, review the details returned in the API error, modify your bootstrap action script, and create a new cluster with the updated bootstrap action.

To troubleshoot the failed EMR cluster, refer to the ErrorDetail information returned from the DescribeCluster and ListClusters APIs. For more information, see Error codes with ErrorDetail information. The ErrorData array within ErrorDetail returns the following information for this error code:


The ID of the primary instance where the bootstrap action failed.


The ordinal number for the bootstrap action that failed. A script with a bootstrap-action value of 1 is the first bootstrap action to run on the instance.


The return code for the bootstrap action that failed.


The Amazon S3 location of the bootstrap action that failed.


The public URL of the documentation for the error code.

Steps to complete

Perform the following steps to identify and fix the root cause of the bootstrap action error. Then launch a new cluster.

  1. Review the bootstrap action log files in Amazon S3 to identify the root cause for the failure. To learn more on how to view Amazon EMR logs, see View log files.

  2. If you turned on cluster logs when you created the instance, refer to the stdout log for more information. You can find the stdout log for the bootstrap action in this Amazon S3 location:


    For more information on cluster logs, see Configure cluster logging and debugging.

  3. To determine the bootstrap action failure, review the exceptions in the stdout logs, and the return-code value in ErrorData.

  4. Use your findings from the previous step to revise your bootstrap action so that it avoids exceptions or can gracefully handle exceptions when they occur.

  5. Launch a new cluster with your updated bootstrap action.