Hello! Has anyone had success modifying the base job template for different flow deployments using the Python sdk? I'm using the worker model with prefect version

2.12.1

on kubernetes and deploy my flows using the

Deployment.build_from_flow()

method. Basically, I'd like to provision more or less mem/cpu based on what I know about each flow deployment. Thanks in advance for any help!

hi @Nathan (nice name btw 🙂 ) can you explain what you mean by

the worker model

? if you're deploying your flows to work pools for workers to pick up, you won't want to use the

build_from_flow

method (an older deployment UX specifically for deployments executed by agents via infra blocks that will eventually be deprecated) as far as kubernetes,

Flow.deploy

should cover the same (declarative python deployment definition) use case for you (which is the recommended approach for worker-based deployments)

Hey Nate (nice name!). Thanks for the tip! I'll look into the

Flow.deploy

method instead. Is there a mechanism to modify the k8s job templates though on a per-deployment basis? I've had a hard time finding that in the docs.

job_variables

contains the

job_manifest

that i assume you'd like to modify here

Awesome. Thank you very much. I'll take a look!

sure thing - one last note, if you want to get a high level view of the structure of the work pool config, the advanced tab on your k8s work pool can be a good visual reference

Nate, is there a way to use

Flow.deploy

without it building a Docker image? I'd rather it just pull an image that's already built on runtime.

Haha nice timing!