if a worker is busy running a flow in kubernetes, and i release a new version of the docker image, what’s to stop kubernetes restarting the worker in order to update it to the latest image? ideally there would be some kind of mechanism to make kubernetes not kill a worker while it’s running a flow, but i’m not clear if there is one?

hey Alexandra, I think it'll depend on how you're deploying your K8s worker and the images it uses to run flows. If you use the helm chart it will by default pull and use the

"2-python3.11-kubernetes"

image. then as long as the worker stays up, it should reuse the same image. in general kubenetes shouldn't be killing the worker -- it should only be restarting when you re-deploy it. are you seeing otherwise?