We are tasked with setting up self-hosted Looker with as much deployment and management automation as possible. We’re considering different deployment models including Kubernetes. But Looker clustering appears to be complex and only superficially documented.
My understanding is that each cluster node will register itself with the cluster via the `--hostname` flag as mentioned briefly in https://docs.looker.com/setup-and-management/tutorials/clustering. Once registered, other nodes will be able to discover the registered node for inter-node communication.
- Can someone help me understand the nature of inter-node communication in Looker clusters? Why do the nodes need to talk to each other at all? What kinds of events cause the nodes to communicate?
- What happens when a node goes away? How does this affect the other nodes? Is there a TTL of some sort in the discovery mechanism so other nodes will eventually give up talking to the disappeared node?