e. Create AWS Batch and Compute Environment

In this step,

  • Create AWS Batch Compute Environment and Job Queue using aws cli

Creating the AWS Batch Compute Environment

The json configuration for the Compute Environment is given below

  • copy the contents to a file called gpu_batch_ce.json and
  • edit the placeholder values
  "computeEnvironmentName": "gpu_mnp_ce",
  "type": "MANAGED",
  "state": "ENABLED",
  "computeResources": {
    "type": "EC2",
    "minvCpus": 0,
    "maxvCpus": 64,
    "desiredvCpus": 0,
    "launchTemplate": {
      "launchTemplateId": "LAUNCH_TEMPLATE",
      "version": "1"
    "instanceTypes": [
    "subnets": [
    "ec2KeyPair": "PEM_KEY_NAME",
    "instanceRole": "INSTANCE_ROLE",
    "tags": {
      "Name": "gpu_mnp_ce"
  "serviceRole": "BATCH_SERVICE_ROLE"

Note: When you are entering strings inside JSON file, it has to be quoted for a valid json

PlaceHolder Replace With
INSTANCE_ROLE arn:aws:iam::xxxxxxxx:instance-profile/ecsInstanceRole
SUBNET subnet-xxxxxxxx
PEM_KEY_NAME key_name w/o .pem extension
BATCH_SERVICE_ROLE arn:aws:iam::xxxxxxxxxxx:role/service-role/AWSBatchServiceRole

After editing the placeholders, execute the command below to create the compute environment

aws batch create-compute-environment --cli-input-json file://gpu_batch_ce.json

Create the AWS Job Queue

The json configuration for the Job Queue ,

  • copy the contents to a file called gpu_batch_jq.json
  • verify the default values which should reflect the names of the compute environment set above
  "jobQueueName": "gpu_mnp_jq",
  "state": "ENABLED",
  "priority": 1,
  "computeEnvironmentOrder": [
      "order": 1,
      "computeEnvironment": "gpu_mnp_ce"
  • Run the aws cli to create the job queue
aws batch create-job-queue --cli-input-json file://gpu_batch_jq.json