What's the easiest way to get a "--from-nodes" "--base-env" sort of experience?

[pascalwhoop]

Often you want to run the entire pipline in env A but start reading the first datasets from env B

E.g. you want to run in Dev but the root datasets should be in prod. When developing we often find ourselves trying to get Prod working and it crashes in K8S/Argo/... Then it's quite tedious to keep working there, you want to move back to your machine. But how do you say

run from node X and read the input data from Prod but then keep going in Dev

? This would be super helpful I think. Right now I copy the datasets over manually from Prod to Dev and then keep workin in Dev. But it would be neat to say --read-from-env prod and this reads the first entrypoints from prod, then keeps going in the env chosen.

[noklam]

@pascalwhoop Maybe I don't understand the question fully. Can you have a duplicate dataset in prod/dev?