Misleading error message for invalid ECS task memory configuration

[rpmccarter]

What happened?

I just bumped the memory for the two containers of my FargateService task from 4GiB + 12GiB to 8GiB + 32GiB. When attempting to deploy this new configuration, I get the following error:

    aws:ecs:TaskDefinition (my-service-prod):
      error: 1 error occurred:
      	* failed creating ECS Task Definition (my-service-prod-718b5d41): ClientException: The 'memory' setting for container 'worker' is greater than for the task.

I could specify the exact memory for the task, but the docstring for taskDefinitionArgs.memory says "If not provided, a default will be computed based on the cumulative needs specified by [containerDefinitions]", so this seems like a bug

This seems very similar to #188

Example

const service = new awsx.ecs.FargateService(`my-service-${stack}`, {
  cluster: cluster.arn,
  assignPublicIp: true,
  desiredCount: 2,
  deploymentMinimumHealthyPercent: 100,
  deploymentMaximumPercent: 200,
  taskDefinitionArgs: {
    taskRole: {
      roleArn: role.arn,
    },
    containers: {
      server: {
        command: ['yarn', 'server'],
        image: image.imageUri,
        cpu: 1024,
        memory: 1024 * 8,
        essential: true,
      },
      worker: {
        command: ['yarn', 'worker'],
        image: image.imageUri,
        cpu: 3072,
        memory: 1024 * 32,
        essential: true,
      },
    },
  },
});

Output of pulumi about

CLI          
Version      3.105.0
Go Version   go1.21.6
Go Compiler  gc

Plugins
NAME        VERSION
aws         5.33.0
awsx        1.0.2
cloudflare  5.0.0
docker      3.6.1
nodejs      unknown
tls         4.10.0

Host     
OS       darwin
Version  14.2.1
Arch     arm64

This project is written in nodejs: executable='/Users/rpmccarter/.nvm/versions/node/v20.10.0/bin/node' version='v20.10.0'

Current Stack: Mintlify/leaves/staging

... (can send if needed, but would prefer to keep private)

Dependencies:
NAME                VERSION
@pulumi/aws         5.33.0
@pulumi/awsx        1.0.2
@pulumi/cloudflare  5.0.0
@pulumi/pulumi      3.60.0
@pulumi/tls         4.10.0
@types/node         16.18.22
rimraf              5.0.5
typescript          5.3.3

Pulumi locates its logs in /var/folders/dn/z0by0dcj1gnbkjr6_t71hp_m0000gn/T/ by default

Contributing

Vote on this issue by adding a 👍 reaction. To contribute a fix for this issue, leave a comment (and link to your pull request, if you've opened one already).

[mjeffryes]

Thanks for the bug report @rpmccarter and pointing out the askDefinitionArgs.memory workaround.

[rpmccarter]

Hi @mjeffryes ! After explicitly adding the 40GiB limit to taskDefinitionArgs.memory, the error changed to:

    aws:ecs:TaskDefinition (leaves-service-dev):
      error: 1 error occurred:
      	* failed creating ECS Task Definition (leaves-service-dev-dbea4070): ClientException: No Fargate configuration exists for given values: 4096 CPU, 40960 memory. See the Amazon ECS documentation for the valid values.

Sure enough, according to the AWS Fargate docs, 4 vCPU and 40GiB is an invalid configuration - the max memory for a 4 vCPU machine is 30 GiB. After decreasing the combined memory of the containers to 30GiB, I was able to remove the explicit task memory limit and everything worked great!

With that said, it's not really a bug, but I suppose the error message could have been a bit clearer - it would have been great if it had specified the invalid configuration initially. I'll update the title of this issue to be more accurate, but feel free to close it if you don't have time to get to it 🙂