This is great. We certainly miss a good Kubernetes chart / operator. One of the pain points people have had though is with rotation. Because of the way the library that we use for Raft works, the cluster either should use static IPs or the pod rotation should be 1 at a time. This causes a lot of problems on k8s when you don't enable proper pod disruption budget. Lately though I've created a

reset-peers-on-error

flag for auto healing clustering, see here: https://github.com/typesense/typesense/issues/465#issuecomment-1487780839