Compute resource AMIs - AWS Batch
Compute resource AMI specification

Compute resource AMIs

By default, AWS Batch managed compute environments use a recent, approved version of the Amazon ECS optimized AMI for compute resources. However, you might want to create your own AMI to use for your managed and unmanaged compute environments. If you require any of the following, we recommend you create your own AMI:

  • Increasing the storage size of your AMI root or data volumes

  • Adding instance storage volumes for supported Amazon EC2 instance types

  • Customizing the Amazon ECS container agent

  • Customizing Docker

  • Configuring a GPU workload AMI to allow containers to access GPU hardware on supported Amazon EC2 instance types

Note

After a compute environment is created, AWS Batch doesn't upgrade the AMIs in the compute environment. AWS Batch also doesn't update the AMIs in your compute environment when a newer version of the Amazon ECS optimized AMI is available. You're responsible for the management of the guest operating system. This includes any updates and security patches. You're also responsible for any additional application software or utilities that you install on the compute resources. To use a new AMI for your AWS Batch jobs, do the following:

  1. Create a new compute environment with the new AMI.

  2. Add the compute environment to an existing job queue.

  3. Remove the earlier compute environment from your job queue.

  4. Delete the earlier compute environment.

In April 2022, AWS Batch added enhanced support for updating compute environments. For more information, see Updating compute environments. To use the enhanced updating of compute environments to update AMIs, follow these rules:

  • Either don't set the service role (serviceRole) parameter or set it to the AWSServiceRoleForBatch service-linked role.

  • Set the allocation strategy (allocationStrategy) parameter to BEST_FIT_PROGRESSIVE or SPOT_CAPACITY_OPTIMIZED.

  • Set the update to latest image version (updateToLatestImageVersion) parameter to true.

  • Don't specify an AMI ID in imageId, imageIdOverride (in ec2Configuration), or in the launch template (launchTemplate). When you don't specify an AMI ID, AWS Batch selects the latest Amazon ECS optimized AMI that AWS Batch supports at the time the infrastructure update is initiated. Alternatively, you can specify the AMI ID in the imageId or imageIdOverride parameters. Or, you can specify the launch template that's identified by the LaunchTemplate properties. Changing any of these properties starts an infrastructure update. If the AMI ID is specified in the launch template, the AMI ID can't be replaced by specifying an AMI ID in either the imageId or imageIdOverride parameters. The AMI ID can only be replaced by specifying a different launch template. If the launch template version is set to $Default or $Latest, the AMI ID can be replaced by setting either a new default version for the launch template (if $Default) or by adding a new version to the launch template (if $Latest).

If these rules are followed, any update that starts an infrastructure update causes the AMI ID to be re-selected. If the version setting in the launch template (launchTemplate) is set to $Latest or $Default, the latest or default version of the launch template is evaluated up at the time of the infrastructure update, even if the launchTemplate wasn't updated.

Topics

Compute resource AMI specification

The basic AWS Batch compute resource AMI specification consists of the following:

Required

  • A modern Linux distribution that's running at least version 3.10 of the Linux kernel on an HVM virtualization type AMI. Windows containers aren't supported.

    Important

    Multi-node parallel jobs can only run on compute resources that were launched on an Amazon Linux instance with the ecs-init package installed. We recommend that you use the default Amazon ECS optimized AMI when you create your compute environment. You can do this by not specifying a custom AMI. For more information, see Multi-node parallel jobs.

  • The Amazon ECS container agent. We recommend that you use the latest version. For more information, see Installing the Amazon ECS Container Agent in the Amazon Elastic Container Service Developer Guide.

  • The awslogs log driver must be specified as an available log driver with the ECS_AVAILABLE_LOGGING_DRIVERS environment variable when the Amazon ECS container agent is started. For more information, see Amazon ECS Container Agent Configuration in the Amazon Elastic Container Service Developer Guide.

  • A Docker daemon that's running at least version 1.9, and any Docker runtime dependencies. For more information, see Check runtime dependencies in the Docker documentation.

    Note

    We recommend the Docker version that ships with and is tested with the corresponding Amazon ECS agent version that you're using. For more information, see Amazon ECS Container Agent Versions in the Amazon Elastic Container Service Developer Guide.

Recommended

  • An initialization and nanny process to run and monitor the Amazon ECS agent. The Amazon ECS optimized AMI uses the ecs-init upstart process, and other operating systems might use systemd. For more information and examples, see Example container instance User Data Configuration Scripts in the Amazon Elastic Container Service Developer Guide. For more information about ecs-init, see the ecs-init project on GitHub. At a minimum, managed compute environments require the Amazon ECS agent to start at boot. If the Amazon ECS agent isn't running on your compute resource, then it can't accept jobs from AWS Batch.

The Amazon ECS optimized AMI is preconfigured with these requirements and recommendations. We recommend that you use the Amazon ECS optimized AMI or an Amazon Linux AMI with the ecs-init package that's installed for your compute resources. Choose another AMI if your application requires a specific operating system or a Docker version that's not yet available in those AMIs. For more information, see Amazon ECS-Optimized AMI in the Amazon Elastic Container Service Developer Guide.