ERR Log "attempted to connect to route port"

[hufan-ola]

In Cluster model, when I pub a message to cluster addr, subscribe can receive message, but we got an ERR log

nats sub foo -s nats://192.168.0.3:6222

[15355] 2022/10/18 11:30:00.638646 [ERR] 192.168.0.3:57824 - rid:10 - attempted to connect to route port

what’s the problem, what should i do ?
Please

config file as below

• main.conf

listen: 192.168.0.3:4222
http: 192.168.0.3:8222
debug: true
trace: false

cluster {
  name: pressure1
  listen: 192.168.0.3:6222
}

• slave.conf

listen: 192.168.0.3:4223
http: 192.168.0.3:8223
debug: true
trace: false

cluster {
  name: pressure1
  listen: 192.168.0.3:6223

  routes = [
    nats-route://192.168.0.3:6222
  ]
}

[Jarema]

Hey!

cluster routes are meant to be used only for other nodes (servers) to connect and form a cluster.

Clients should connect to the listen URL of one of the nodes. In your example it's 192.168.0.3:4222.

Docs are here: https://docs.nats.io/running-a-nats-service/configuration/clustering

Keep in mind that your cluster should have at least 3 nodes.
If you want master/slave hub/spoke architecture, leafnodes might be a better solution.

[hufan-ola]

SubA listen for msg

nats -s nats://192.168.0.243:4222 sub foo

SubA can received msg below three port for publish

nats -s nats://192.168.0.243:4222 pub foo h1
nats -s nats://192.168.0.243:4223 pub foo h2
nats -s nats://192.168.0.243:4224 pub foo h3

But if Seed Node crashed, SubA can attempt reconnect to port 4223 automatic

Disconnected due to: EOF, will attempt reconnect
Reconnected [nats://192.168.0.243:4223]

but client will report an error as below

nats -s nats://192.168.0.243:4222 pub foo h1
nats: error: nats: no servers available for connection

What I need is client can use one address to connect the Cluster, and do not need appoint specific address(eg: 192.168.0.243:4222) and client can publish success as long as any node in Cluster is alive

Can we did it? @Jarema

[Jarema]

When connecting to any of the nodes, Client will get information on other nodes in the cluster. That's why your subscribe example works by automatically reconnecting.

If you need to be able to connect to the cluster, no matter which servers are up, using only one URL, that's out of the scope of servers setup.
You probably need something like a DNS with multiple entries, or something relevant to your setup.

You can also pass more URLs to the connect nats pub foo data --server localhost:4222,localhost:4223 and client will connect to the one that is available.