EMQX cluster with core and replicant nodes


the documentation suggests to use a cluster that consists of core and replicant nodes when aiming for millions of connections.

We are already running a cluster with core nodes only on AWS ECS. We want to create another AWS ECS cluster that runs the replicant nodes now.

  • Is it best practice to have run two cluster with core and replicant nodes each?
  • For the replicant nodes: is it possible to specify the core nodes in the config like this?
core_nodes = "cluster.emqx.local" (DNS A record that holds all core node IPs)

instead of

core_nodes = "emqx1@, emqx2@ ..."


Yes, it’s highly recommended to use domain names rather than IP addresses for the node names. Also, setting the core_nodes configuration variable is now optional. Replicant nodes will discover the core nodes using the same cluster discovery mechanism as per usual.