Opensearch

It's built for the KubeSphere observability platform and comes with master nodes for coordination, data nodes for storage, an optional dashboard for visualization, and a curator for cleaning old indices. The documentation is thorough about cluster selection, which matters since you're dealing with multi-cluster deployments. Default setup uses NodePort 30920 for external access, and it's opinionated about making you confirm target clusters before proceeding. If you're running KubeSphere and need centralized search and analytics for observability data, this gets you there with reasonable defaults and clear configuration options for scaling the data nodes or tweaking JVM memory.