Hi everyone, I am new to dagster and Kubernetes. I set up my dagster following this link . I want to run every op of a graph on different pods. I tried with K8sRunLauncher (with kubernetes_executor), and everything works fine. But the problem with this launcher is that every time I trigger a dag, it creates new jobs and pods. When I trigger it so many times, it clutters up the cluster. I get a list of 1000+ pods (which is very inefficient). I found one solution (TTL controller link ) for the automatic clearing of successful or failed pods. But I want a more efficient solution like reusing pods defined for dag. So I have some questions: 1. Is it possible to define a pod for every op in a graph and reuse the same pod whenever I trigger a dag? 2. If it's possible in dagster, can someone share any blog or dagster documentation page regarding the implementation of this kind of dag?

Hey Anup - we don't have a feature right now that lets you run each op in its own standing server currently unfortunately. We do have a run launcher that lets you launch runs against the user code deployment's server - but nothing that works at the op level in the way it sounds like you're looking for

Daniel, thank you for replying. I will check how can I adjust according to your platform. Dagster provide much better features compare to other tools.