Changes AWS Batch adds support for Amazon EFS File System
Changes This release adds support for customer to run Batch Jobs on ECS Fargate, the serverless compute engine built for containers on AWS. Customer can also propagate Job and Job Definition Tags to ECS Task.
Changes Add Ec2Configuration in ComputeEnvironment.ComputeResources. Use in CreateComputeEnvironment API to enable AmazonLinux2 support.
Changes Adding evaluateOnExit to job retry strategies.
Changes Support tagging for Batch resources (compute environment, job queue, job definition and job) and tag based access control on Batch APIs
Changes Support custom logging, executionRole, secrets, and linuxParameters (initProcessEnabled, maxSwap, swappiness, sharedMemorySize, and tmpfs). Also, add new context keys for awslogs.
Changes Adding support for Compute Environment Allocation Strategies
Changes AWS Batch now supports SDK auto-pagination and Job-level docker devices.
Changes Support for GPU resource requirement in RegisterJobDefinition and SubmitJob
Changes Adding multinode parallel jobs, placement group support for compute environments.
Changes Adding EC2 Launch Template support in AWS Batch Compute Environments.
Changes Support for Timeout in SubmitJob and RegisterJobDefinition
Changes Add support for Array Jobs which allow users to easily submit many copies of a job with a single API call. This change also enhances the job dependency model to support N_TO_N and sequential dependency chains. The ListJobs and DescribeJobs APIs now have the ability to list or describe the status of entire Array Jobs or individual elements within the array.
Changes This release enhances the DescribeJobs API to include the CloudWatch logStreamName attribute in ContainerDetail and ContainerDetailAttempt
Changes API Update for AWS Batch: Customer provided AMI for MANAGED Compute Environment
Changes Customers can now provide a retryStrategy as part of the RegisterJobDefinition and SubmitJob API calls.
Changes AWS Batch is a batch computing service that lets customers define queues and compute environments and then submit work as batch jobs.