While investigating an outage on our production cluster, I noticed that if I turn off one of the data nodes, it kills the entire cluster. The cluster immediately becomes operational when I turn it back on.
What is going on? How can I further investigate?
5 Data Nodes
All have X-Pack Platinum
Theory.... Is it possible that the node has an index on it that is needed for x-pack and when it goes down it brings the cluster down with it?