I'm looking for a way to automate the installation of MLOps tools like Kubeflow and MLflow, as well as services like Spark and Airflow, whenever I set up a new Kubernetes cluster. Right now, I handle the installation manually using Juju and Helm, but it really slows down my testing process. I'm thinking about using Kubebuilder to create a custom operator to manage this, but I've run into some issues with Juju. Does anyone have any suggestions or experiences that might help?
1 Answer
Skip using Juju and check out tools like Flux or ArgoCD with GitOps. It's simpler than creating a custom operator, and this problem of automating installations has already been figured out. Just leverage the existing solutions instead!
Can you share a specific example of how you would implement that? I'm particularly interested in automating the setup through an API when a new cluster is spun up.