Could someone just check that I understand correctly: • If I use the

K8sRunLauncher

as outlined in deploying-with-helm , then each pipeline gets executed in a Job using the same container as the User Code Deployment. • If I use the

CeleryK8sRunLauncher

from deploying-with-helm-advanced then each solid gets executed in its own Job (again using the same container as the User Code Deployment). Is this all correct?

Yes, just some nits: The second bullet requires the CeleryK8sRunLauncher and corresponding CeleryK8sExecutor (attached to pipeline modes). The run launcher kicks off a process which traverses the pipeline dag, while the executor handles running the actual solid. The K8sRunLauncher is generally used in tandem with either the inprocess or multiprocess executors, which as you describe run the pipeline within the k8s pod

again using the same container image as the User Code Deployment)

please! I love the tiny details when learning

Edited 🙂

thank you!