fix: always monitor manager

lib/supavisor/client_handler.ex

@@ -395,10 +395,10 @@ defmodule Supavisor.ClientHandler do
              else: true
            ),
          {:ok, opts} <- Supavisor.subscribe(sup, data.id) do
-      Process.monitor(opts.workers.manager)
+      manager_ref = Process.monitor(opts.workers.manager)
       data = Map.merge(data, opts.workers)
       db_pid = db_checkout(:both, :on_connect, data)
-      data = %{data | db_pid: db_pid, idle_timeout: opts.idle_timeout}
+      data = %{data | manager: manager_ref, db_pid: db_pid, idle_timeout: opts.idle_timeout}
 
       next =
         if opts.ps == [],
@@ -424,6 +424,8 @@ defmodule Supavisor.ClientHandler do
       :proxy ->
         case Supavisor.get_pool_ranch(data.id) do
           {:ok, %{port: port, host: host}} ->
+            manager_ref = Process.monitor(opts.workers.manager)
+
             auth =
               Map.merge(data.auth, %{
                 port: port,
@@ -434,7 +436,8 @@ defmodule Supavisor.ClientHandler do
                 upstream_verify: nil
               })
 
-            {:keep_state, %{data | auth: auth}, {:next_event, :internal, :connect_db}}
+            {:keep_state, %{data | auth: auth, manager: manager_ref},
+             {:next_event, :internal, :connect_db}}
 
           other ->
             Logger.error("ClientHandler: Subscribe proxy error: #{inspect(other)}")
@@ -623,7 +626,8 @@ defmodule Supavisor.ClientHandler do
   end
 
   # pool's manager went down
-  def handle_event(:info, {:DOWN, _, _, _, reason}, state, data) do
+  def handle_event(:info, {:DOWN, ref, _, _, reason}, state, %{manager: manager_ref} = data)
+      when ref == manager_ref do
     Logger.error(
       "ClientHandler: Manager #{inspect(data.manager)} went down #{inspect(reason)} state #{inspect(state)}"
     )