reader closed during delivery tests

[h0jeZvgoxFepBQ2C]

Right now we are testing nats.ws and it occured to us, that when getting flooded with bigger messages (300 messages x 400 kilobytes), following error happens:

This doesn't happen when we send the same content, just less characters.
After this error, the client is not receiving anymore.

Also the connection monitoring doesn't work, there is no disconnection event triggered, so the client also doesn't reconnect?

This is our code:

  const connectionOptions = {
    servers: ["ws://localhost:4333"],
    debug: true,
    maxReconnectAttempts: -1,
    reconnectTimeWait: 2000,
    timeout: 3000
  }

useEffect(() => {
    setConnectionState("connecting")

    const connectAndSubscribe = async () => {
      try {
        const nc = await connect(connectionOptions)
        setConnectionState("connected")
        console.info(`connected ${nc.getServer()}`);

        channels.map(channelName => {
          const sub = nc.subscribe(channelName);
          console.log("Subscribing to ", channelName);
          (async () => {
            for await (const m of sub) {
              const sc = StringCodec();
              try {
                const result = JSON.parse(sc.decode(m.data))
                onMessageCompleted("nats", result.id, result)
              } catch(ex) {
                console.error(ex)
              }
            }
            console.log("subscription closed");
          })();
        });

        (async () => {
          for await (const s of nc.status()) {
            console.info("mmmmm", s, s.type);
            switch(s.type) {
            case "disconnect":
              setConnectionState("disconnected");
              break;
            case "reconnect":
              setConnectionState("connected");
              break;
            case "reconnecting":
              console.log("reconnectingreconnectingreconnectingreconnectingreconnecting", connectionRetries);
              setConnectionRetries(connectionRetries + 1);
              setConnectionState("connecting");
              break;
            case "error":
              setConnectionState("failed");
              break;
            }
          }
        })();

        nc.closed().then((err) => {
          if (err) {
            console.log('Connection error:', err.message);
            handleConnectionError();
          } else {
            console.log('Connection closed');
          }
        });

        window.nats = nc

      } catch (err) {
        console.log('Failed to connect:', err.message);
        handleConnectionError()
      }
    }

    const handleConnectionError = () => {
      setTimeout(() => {
        connectAndSubscribe();
      }, connectionOptions.reconnectTimeWait);
    }
    connectAndSubscribe()

    return () => {
      if (window.nats) {
        window.nats.close();
      }
    };
  }, [])

[aricart]

@h0jeZvgoxFepBQ2C what server version are you using?

[h0jeZvgoxFepBQ2C]

Version: 2.9.16 run via Docker desktop.

Here is our config file:

# Client port of 4222 on all interfaces
port: 4222

# HTTP monitoring port
monitor_port: 8222
debug:   true
max_payload: 64MB

# This is for clustering multiple servers together.
cluster {
  # It is recommended to set a cluster name
  name: "my_cluster"

  # Route connections to be received on any interface on port 6222
  port: 6222

  # Routes are protected, so need to use them with --routes flag
  # e.g. --routes=nats-route://ruser:T0pS3cr3t@otherdockerhost:6222
  authorization {
    user: ruser
    password: T0pS3cr3t
    timeout: 2
  }

  # Routes are actively solicited and connected to from this server.
  # This Docker image has none by default, but you can pass a
  # flag to the nats-server docker image to create one to an existing server.
  routes = []
}

websocket {
  port: 4333
  no_tls: true,
}

and here our docker-compose entry:

  nats:
    image: nats
    ports:
      - "4222:4222"
      - "4333:4333"
      - "8222:8222"
    volumes:
        - ./cluster/dev/nats/nats.conf:/nats-server.conf

[aricart]

go back to 2.9.15 - I think there's an issue on the server, need to investigate.

[aricart]

Can you give me more info on what your code is doing, need a test to reproduce.

[h0jeZvgoxFepBQ2C]

Yep thanks, this was the issue. 2.9.15 works fine. Will give you some more code in some minutes.

[h0jeZvgoxFepBQ2C]

We are using the ruby gem for sending and a react app for receiving (code above):

    size = 4.kilobytes # or 400.kilobytes, which didn't work in 2.9.16
    300.times do |i|
      msg = { i: i, type: event, data: message}.to_json

      nats = NATS.connect({
        servers: ["nats://127.0.0.1:4222"],
        connect_timeout: 30, 
        write_timeout: 30, 
        read_timeout: 30
      })
      nats.publish("test", msg)
      nats.drain
    end

[h0jeZvgoxFepBQ2C]

But with 2.9.15 everything works fine, thank you so much, also for the ultra fast response ❤️ ❤️ ❤️

[h0jeZvgoxFepBQ2C]

Shall I leave this issue open until this gets fixed in the next version or shall I close it?

[aricart]

@h0jeZvgoxFepBQ2C server group is looking at it

[h0jeZvgoxFepBQ2C]

Thank you so much ❤️