fix: observer load command crash when loading noexist module

apps/emqx_modules/src/emqx_modules.app.src

@@ -1,7 +1,7 @@
 %% -*- mode: erlang -*-
 {application, emqx_modules, [
     {description, "EMQX Modules"},
-    {vsn, "5.0.22"},
+    {vsn, "5.0.23"},
     {modules, []},
     {applications, [kernel, stdlib, emqx, emqx_ctl]},
     {mod, {emqx_modules_app, []}},

apps/emqx_modules/src/emqx_observer_cli.erl

@@ -40,10 +40,18 @@ cmd(["bin_leak"]) ->
         recon:bin_leak(100)
     );
 cmd(["load", Mod]) ->
-    Module = list_to_existing_atom(Mod),
-    Nodes = nodes(),
-    Res = remote_load(Nodes, Module),
-    emqx_ctl:print("Loaded ~p module on ~p: ~p~n", [Module, Nodes, Res]);
+    case nodes() of
+        [] ->
+            emqx_ctl:print("No other nodes in the cluster~n");
+        Nodes ->
+            case emqx_utils:safe_to_existing_atom(Mod) of
+                {ok, Module} ->
+                    Res = recon:remote_load(Nodes, Module),
+                    emqx_ctl:print("Loaded ~p module on ~p: ~p~n", [Module, Nodes, Res]);
+                {error, Reason} ->
+                    emqx_ctl:print("Module(~s) not found: ~p~n", [Mod, Reason])
+            end
+    end;
 cmd(_) ->
     emqx_ctl:usage([
         {"observer status", "Start observer in the current console"},
@@ -51,12 +59,5 @@ cmd(_) ->
             "Force all processes to perform garbage collection "
             "and prints the top-100 processes that freed the "
             "biggest amount of binaries, potentially highlighting leaks."},
-        {"observer load Mod", "Ensure a module is loaded in all EMQX nodes in the cluster"}
+        {"observer load Mod", "Enhanced module synchronization across all cluster nodes"}
     ]).
-
-%% recon:remote_load/1 has a bug, when nodes() returns [], it is
-%% taken by recon as a node name.
-%% before OTP 23, the call returns a 'badrpc' tuple
-%% after OTP 23, it crashes with 'badarg' error
-remote_load([], _Module) -> ok;
-remote_load(Nodes, Module) -> recon:remote_load(Nodes, Module).