Setting up Dask and Ray on Armada is more work than on K8s due to not being able to use published helm charts to deploy clusters. Being able to do this would make testing and integrating new distributed frameworks much faster.
┆Issue is synchronized with this Jira Task by Unito
