Description
Currently Elasticsearch allows starting multiple nodes using the same data path by simply starting multiple processes. This is great for the OOB experience where people can simply build clusters without too much hassle. On the other side, this has caused much confusion in real clusters where people started a node before properly shutting down the first, which looks to the outside as data loss. There are valid production use cases where we want multiple nodes on the same servers but those are rare (and mostly involve running on servers with lots of memory).
I think it is a good idea to better protect people by using our new "production mode" definition to change the default of the node.max_local_storage_nodes to 1, from the currently unbounded value.