aws-batch: MultiNodeJobDefinition Incorrectly Requires InstanceType

[SeanStockwell]

Describe the bug

It seems that the Batch MultiNodeJobDefinition CDK Construct requires instance type to be defined. However, instanceType is not required by the Batch AWS SDK.

This seems to be more than a documentation issue, as I'm unable to create a MultiNodeJobDefinition in my CDK package without specifying instance type.

This constraint takes away from one of the primary advantages to having Batch manage our EC2 resources in that they optimally assign batch jobs to instance types based on availability/resource requirements.

Expected Behavior

We shouldn't need to specify instance type. Batch should determine it based on job resource requirements.

Current Behavior

CDK requires instance type to be required at job definition level.

Reproduction Steps

Create a MultiNodeJobDefinition.

Possible Solution

No response

Additional Information/Context

No response

CDK CLI Version

2.96.2

Framework Version

No response

Node.js Version

18.x

OS

linux

Language

Typescript

Language Version

No response

Other information

No response

[indrora]

This is because multinode jobs (through a quirk of CloudFormation) requires it.

https://docs.aws.amazon.com/AWSCloudFormation/latest/UserGuide/aws-properties-batch-jobdefinition-containerproperties.html#cfn-batch-jobdefinition-containerproperties-instancetype

[SeanStockwell]

The link you provided says that InstanceType is not required

[peterwoodworth]

Maybe @comcalvi can comment on why this is required in our construct, but as far as I can tell it isn't required in CloudFormation. I implemented this aspect on my stack and successfully deployed.

class BatchContainerDefinitionModifier implements cdk.IAspect {
  public visit(node: IConstruct): void {
    // will need to be stricter in search if you are creating other job definitions
    if (node instanceof batch.CfnJobDefinition) {
      node.addPropertyDeletionOverride('NodeProperties.NodeRangeProperties.0.Container.InstanceType')
      node.addPropertyDeletionOverride('NodeProperties.NodeRangeProperties.1.Container.InstanceType')
      // repeat for as many containers as necessary
    }
  }
}

cdk.Aspects.of(stack).add(new BatchContainerDefinitionModifier());

repro stack

    const def = new batch.MultiNodeJobDefinition(this, 'JobDefinition', {
      instanceType: new ec2.InstanceType('m5.large'),
    });

    def.addContainer({
      startNode: 0,
      endNode: 5,
      container: new batch.EcsEc2ContainerDefinition(this, 'multiContainer', {
        image: ecs.ContainerImage.fromRegistry('amazon/amazon-ecs-sample'),
        cpu: 256,
        memory: cdk.Size.mebibytes(2048),
      })
    });

    def.addContainer({
      startNode: 6,
      endNode: 10,
      container: new batch.EcsEc2ContainerDefinition(this, 'multiContainer2', {
        image: ecs.ContainerImage.fromRegistry('amazon/amazon-ecs-sample'),
        cpu: 256,
        memory: cdk.Size.mebibytes(2048),
      })
    });

[github-actions]

⚠️COMMENT VISIBILITY WARNING⚠️

Comments on closed issues are hard for our team to see.
If you need more assistance, please either tag a team member or open a new issue that references this one.
If you wish to keep having a conversation with other community members under this issue feel free to do so.