I have the following supervisor children:

       [
          %{
            id: :ranch_sup,
            start: {:ranch_sup, :start_link, []},
            restart: :permanent,
            shutdown: 5000,
            type: :supervisor
          },
          :ranch.child_spec(Server, :ranch_tcp, [port: 6666], Server, [])
        ]

and in mix.exs:

  # Run "mix help compile.app" to learn about applications.
  def application do
    [
      extra_applications: [:logger],
      included_applications: [:ranch],
      mod: {Server.Application, []}
    ]
  end

  # Run "mix help deps" to learn about dependencies.
  defp deps do
    [
      {:ranch, "~> 1.7", runtime: false}
    ]
  end

The problem is that elixir makes :ranch start its supervisor automatically, I guess … which causes

18:00:50.418 [info]  Application server exited: Server.Application.start(:normal, []) returned an error: shutdown: failed to start child: :ranch_sup
    ** (EXIT) already started: #PID<0.895.0>
{"Kernel pid terminated",application_controller,"{application_start_failure,server,{{shutdown,{failed_to_start_child,ranch_sup,{already_started,<0.895.0>}}},{'Elixir.Server.Application',start,[normal,[]]}}}"}
Kernel pid terminated (application_controller) ({application_start_failure,server,{{shutdown,{failed_to_start_child,ranch_sup,{already_started,<0.895.0>}}},{'Elixir.Server.Application',start,

Crash dump is being written to: erl_crash.dump...done