Gateway.* settings and cluster scale-out

I'm reading here that gateway.* settings take effect only after a full cluster restart.

Now I want to add another data node to a cluster and thus I want to bump gateway.expected_data_nodes and gateway.recover_after_data_nodes. So:

  • I have to distribute this settings to elasticsearch.yaml of every node in the cluster, right?
  • Do I have to actually restart the nodes or will they pick up this setting when coming up after next potential cluster shutdown?

Anyone please?