AMQP 连接丢失不会终止 parent 进程,因此永远不会重新连接

AMQP connection lost doesn't kill parent process, so reconnection never happens

我的 GenServer 有初始化功能。 Supervisor 正在照看它,应该在退出时重新启动它。

  def init(_opts) do
    username = get_conf(:username)
    password = get_conf(:password)
    host = get_conf(:host)
    port = get_conf(:port)
    vhost = String.replace(get_conf(:vhost), "/", "%2f")
    {:ok, conn} = Connection.open("amqp://#{username}:#{password}@#{host}:#{port}/#{vhost}")
    {:ok, chan} = Channel.open(conn)
    state = %State{
      exchange: get_conf(:exchange),
      channel: chan,
      routing_key: get_conf(:routing_key)
    }
    {:ok, state}
  end

当我重新启动 RabbitMQ 时 sudo service rabbitmq-server restart 未建立新连接。

在调试栏中,我看到以下内容: 当我单击连接 pid <0.417.0> 时,我收到消息说进程不再存在。似乎进程已死,parent AmqpTransport 对此一无所知。 我怎样才能让AmqpTransport和它的child一起死Connection

通过添加 trapping exits 并链接到 Connection 进程修复了它。此外,我在连接期间出现模式匹配错误以避免 reached_max_restart_intensity.

  @restart_delay 2000 # 2 seconds

  def init(_opts) do
    Process.flag(:trap_exit, true)
    send(self(), :connect)
    {:ok, nil}
  end

  def handle_info(:connect, state) do
    #...
    case Connection.open("amqp://#{username}:#{password}@#{host}:#{port}/#{vhost}") do
      {:ok, conn} ->
        Process.link(conn.pid)
        # ...
        {:noreply, state}
      {:error, :econnrefused} ->
        Logger.error("amqp transport failed with connection refused")
        Process.send_after(self(), :connect, @restart_delay)
        {:noreply, nil}
    end

  end

  def handle_info({:EXIT, pid, reason}, state) do
    Logger.error("amqp transport failed with #{inspect(reason)}")
    Process.unlink(pid)
    Process.send_after(self(), :connect, @restart_delay)
    {:noreply, nil}
  end