Instance type not supported

If you create a cluster, and it fails with the error message "The requested instance type InstanceType is not supported in the requested Availability Zone," it means that you created the cluster and specified an instance type for one or more instance groups that is not supported by Amazon EMR in the Region and Availability Zone where the cluster was created. Amazon EMR may support an instance type in one Availability Zone within a Region and not another. The subnet you select for a cluster determines the Availability Zone within the Region.

Solution

Determine available instance types in an Availability Zone using the AWS CLI

• Use the ec2 run-instances command with the --dry-run option. In the example below, replace m5.xlarge with the instance type you want to use, ami-035be7bafff33b6b6 with the AMI associated with that instance type, and subnet-12ab3c45 with a subnet in the Availability Zone you want to query.

  aws ec2 run-instances --instance-type m5.xlarge --dry-run --image-id ami-035be7bafff33b6b6 --subnet-id subnet-12ab3c45

  For instructions on finding an AMI ID, see Find a Linux AMI. To find a subnet ID, you can use the describe-subnets command.

To learn more about how to discover available instance types, see Find an Amazon EC2 instance type.

After you determine the instance types available, you can do any of the following:

• Create the cluster in the same Region and EC2 Subnet, and choose a different instance type with similar capabilities as your initial choice. For a list of supported instance types, see Supported instance types with Amazon EMR. To compare capabilities of EC2 instance types, see Amazon EC2 instance types.

• Choose a subnet for the cluster in an Availability Zone where the instance type is available and supported by Amazon EMR.

Mitigate instance fleet cluster launch failures because of unsupported primary instance types in Amazon EMR

Primary nodes are essential in Amazon EMR clusters. An EMR cluster launch might fail with an instance type not supported error where Amazon EMR attempts to launch the cluster in an Availability Zone where if the primary instance type isn't supported. Enhanced Availability Zone selection for instance fleet clusters in Amazon EMR automatically filters out unsupported AZs for the primary instance types you specified in the cluster configuration. This means Amazon EMR won't choose an Availability Zone where the configured primary instance types are not supported, which prevents cluster launch failures because of unsupported instance types.

To enable this improvement, add the required permission to the service role or policy for your cluster. The latest version of AmazonEMRServicePolicy_v2 includes this permission, so if you use that policy, the improvement is already available. If you use a custom service role or policy, add the permission ec2:DescribeInstanceTypeOfferings when launching your cluster.

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Action": "ec2:DescribeInstanceTypeOfferings",
            "Effect": "Allow",
            "Resource": "*"
        }
    ]
}