What’s the recommended way of launching a flow that branches out to multiple nodes on kubernetes?

hi @Abhinav Chordia - i think there's a couple ways you could do that. exact recommendation i think would depend on the specifics of your use case it seems you could: • have independent deployments using a kubernetes work pool with deployment-level job spec overrides (specifying nodeSelector or something) and trigger runs of these deployments with

run_deployment

from a central orchestrator flow (for example, here the

image

and

cluster_config

are being overriden for any deployment created with that

prefect.yaml

). you can go into workpools > your workpool > edit > base job template > advanced, to see whats available to override • kick off work yourself wherever you need to from an orchestrating flow with the kubernetes client

To add onto that - what are your conditions for multiple nodes ? Are there labels and taints applied ? Do you want to just maximize usage, or guide pods to specific node pools?

I’d like to maximize usage to start. Scheduling for specific node pools is a nice to have for more production-ability for larger workloads