Hey, happy new year! I'm configuring an ECS work pool. I've got flows running successfully on an ECS Fargate cluster, but want to specify

cpu/memory

allocations on a per-flow basis. Right now, the implementation includes • An ECS cluster with an ECS worker running on it • A single Docker image containing all code for all flows • A base job template on the ECS work pool that specifies the cluster and image The ECS worker spawns flow runs on the cluster, but it always spawns flow runs in containers with the default

1024/2048

cpu/memory

allocations I've been trying to override these per-flow using

Deployment.build_from_flow()

. See the attached example. The commented out sections are the ways I've tried overriding the

cpu/memory

allocations, but none have worked. I'd prefer to not have to build each individual flow into separate docker images. We've got close to 100 flows that all run out of the same codebase, so having a single image seems the sensible way to do things. I feel like I'm missing something crucial. Any pointers would be much appreciated.

the

infra_overrides

should be what you want here, you're "overriding" the value on a work pool for a specific deployment (without changing the work pool directly) can you explain whats not working in that regard? what does the Configuration tab show in the UI after building a deployment with an

infra_overrides

?

@Nate that worked. I thought I was going crazy. Thank you very much!

np, glad you got it working catjam