Hi all — I have a use case where I want to execute a run in my dev environment, where part of the flow reads artifacts from a previously existing run that was executed in my stage environment. What is the recommended way to achieve this? I can use the global

metadata()

function, but it leads to some surprising behavior (see screenshot) and can be dangerous/error-prone.

@cool-father-88039 - is my interpretation correct that the staging stack is an entirely different metaflow stack?

No, it's basically a copy. Just different environments with AWS and kubernetes. It's a normal dev/stage/prod split common in software engineering.

there are a few different ways of doing the dev/stage/prod split - one extreme would be complete isolation (different accounts, clusters, dbs etc.) and another would be one single infrastructure stack but relying on metaflow tags to logically distinguish between dev/stage/prod workloads

it's different accounts, clusters, dbs