In this example, because you’re creating a new Flow Run, it would spin up a separate piece of infrastructure (defined in your Run Config) for that Flow run. Let’s say we’re running on Kubernetes and we’re using a DaskExecutor configured for K8s.
* Flow A runs, realizes it has a custom task to run
* Flow A kicks off one Flow Run of Flow B
* Agent creates new K8s job for the Flow Run of Flow B
* Flow Run of Flow B creates its own executor infrastructure using the DaskExecutor, runs, finishes
* The Flow Run of Flow A continues