Hello! I have Elasticsearch running in Kubernetes, and there is a cluster in each AWS Availability Zone (AZ). I am working on setting up cross-cluster searching across all AZs and currently have it working, but not in a desirable/future-proof way.
In each K8s cluster, I set up a NodePort for Elasticsearch, and then my elasticsearch.yml looks something like this (elasticsearch.yml in AZ A hitting AZ B):
--- cluster.name: "docker-cluster" network.host: 0.0.0.0 discovery.zen.minimum_master_nodes: 1 search: remote: cluster_a: seeds: "127.0.0.1:9300" cluster_b: seeds: "<IP of worker node in AZ B>:<NodePort>"
The problem with this is the NodePort will change in the event of a redeployment, and the node IP will change if a new cluster is deployed. I have an ingress and Application Load Balancer (ALB) for all applications and would like to know if it was possible to either use the ingress and specify a path (as other apps are using the ingress) or - ideally - to use the hostname used by the ALB and the path. Hostname doesn't seem to be a problem, but adding a path doesn't seem to work. If that doesn't work, is there a better recommended method for achieving this?