Merge pull request #9210 from thalesmg/fix-mgmt-endpoint-status-code-v43

fix(mgmt_api): return 503 when emqx is not running in `/status` (v4.3)

apps/emqx_management/src/emqx_management.app.src

@@ -1,6 +1,6 @@
 {application, emqx_management,
  [{description, "EMQ X Management API and CLI"},
-  {vsn, "4.3.18"}, % strict semver, bump manually!
+  {vsn, "4.3.19"}, % strict semver, bump manually!
   {modules, []},
   {registered, [emqx_management_sup]},
   {applications, [kernel,stdlib,minirest]},

apps/emqx_management/src/emqx_mgmt_http.erl

@@ -114,7 +114,11 @@ handle_request(<<"GET">>, <<"/status">>, Req) ->
     end,
     Status = io_lib:format("Node ~s is ~s~nemqx is ~s",
                             [node(), InternalStatus, AppStatus]),
-    cowboy_req:reply(200, #{<<"content-type">> => <<"text/plain">>}, Status, Req);
+    StatusCode = case AppStatus of
+                     running -> 200;
+                     not_running -> 503
+                 end,
+    cowboy_req:reply(StatusCode, #{<<"content-type">> => <<"text/plain">>}, Status, Req);
 
 handle_request(_Method, _Path, Req) ->
     cowboy_req:reply(400, #{<<"content-type">> => <<"text/plain">>}, <<"Not found.">>, Req).

apps/emqx_management/test/emqx_mgmt_api_SUITE.erl

@@ -25,6 +25,8 @@
 -include_lib("emqx/include/emqx_mqtt.hrl").
 -include_lib("emqx_management/include/emqx_mgmt.hrl").
 
+-define(HOST, "http://127.0.0.1:8081/").
+
 -import(emqx_mgmt_api_test_helpers,
         [request_api/3,
          request_api/4,
@@ -44,9 +46,31 @@ end_per_suite(Config) ->
     emqx_ct_helpers:stop_apps([emqx_management]),
     Config.
 
+init_per_testcase(t_status_ok, Config) ->
+    ok = emqx_rule_registry:mnesia(boot),
+    ok = emqx_dashboard_admin:mnesia(boot),
+    application:ensure_all_started(emqx_rule_engine),
+    application:ensure_all_started(emqx_dashboard),
+    Config;
+init_per_testcase(t_status_not_ok, Config) ->
+    ok = emqx_rule_registry:mnesia(boot),
+    ok = emqx_dashboard_admin:mnesia(boot),
+    application:ensure_all_started(emqx_rule_engine),
+    application:ensure_all_started(emqx_dashboard),
+    application:stop(emqx),
+    Config;
 init_per_testcase(_, Config) ->
     Config.
 
+end_per_testcase(t_status_ok, _Config) ->
+    application:stop(emqx_rule_engine),
+    application:stop(emqx_dashboard),
+    ok;
+end_per_testcase(t_status_not_ok, _Config) ->
+    application:stop(emqx_rule_engine),
+    application:stop(emqx_dashboard),
+    application:ensure_all_started(emqx),
+    ok;
 end_per_testcase(_, Config) ->
     Config.
 
@@ -656,6 +680,51 @@ t_data_import_content(_) ->
     application:stop(emqx_rule_engine),
     application:stop(emqx_dashboard).
 
+t_status_ok(_Config) ->
+    {ok, #{ body := Resp
+          , status_code := StatusCode
+          }} = do_request(#{method => get, path => ["status"], headers => [],
+                            body => no_body}),
+    ?assertMatch(
+      {match, _},
+      re:run(Resp, <<"emqx is running$">>)),
+    ?assertEqual(200, StatusCode),
+    ok.
+
+t_status_not_ok(_Config) ->
+    {ok, #{ body := Resp
+          , status_code := StatusCode
+          }} = do_request(#{method => get, path => ["status"], headers => [],
+                            body => no_body}),
+    ?assertMatch(
+      {match, _},
+      re:run(Resp, <<"emqx is not_running$">>)),
+    ?assertEqual(503, StatusCode),
+    ok.
+
+do_request(Opts) ->
+    #{ path := Path
+     , method := Method
+     , headers := Headers
+     , body := Body0
+     } = Opts,
+    URL = ?HOST ++ filename:join(Path),
+    Request = case Body0 of
+                  no_body -> {URL, Headers};
+                  {Encoding, Body} -> {URL, Headers, Encoding, Body}
+              end,
+    ct:pal("Method: ~p, Request: ~p", [Method, Request]),
+    case httpc:request(Method, Request, [], []) of
+        {error, socket_closed_remotely} ->
+            {error, socket_closed_remotely};
+        {ok, {{_, StatusCode, _}, Headers1, Body1}} ->
+            Body2 = case emqx_json:safe_decode(Body1, [return_maps]) of
+                        {ok, Json} -> Json;
+                        {error, _} -> Body1
+                    end,
+            {ok, #{status_code => StatusCode, headers => Headers1, body => Body2}}
+    end.
+
 filter(List, Key, Value) ->
     lists:filter(fun(Item) ->
         maps:get(Key, Item) == Value

apps/emqx_management/test/emqx_mgmt_api_test_helpers.erl

@@ -36,13 +36,21 @@ request_api(Method, Url, QueryParams, Auth, []) ->
                  "" -> Url;
                  _ -> Url ++ "?" ++ QueryParams
              end,
-    do_request_api(Method, {NewUrl, [Auth]});
+    Headers = case Auth of
+                  no_auth -> [];
+                  Header  -> [Header]
+              end,
+    do_request_api(Method, {NewUrl, Headers});
 request_api(Method, Url, QueryParams, Auth, Body) ->
     NewUrl = case QueryParams of
                  "" -> Url;
                  _ -> Url ++ "?" ++ QueryParams
              end,
-    do_request_api(Method, {NewUrl, [Auth], "application/json", emqx_json:encode(Body)}).
+    Headers = case Auth of
+                  no_auth -> [];
+                  Header  -> [Header]
+              end,
+    do_request_api(Method, {NewUrl, Headers, "application/json", emqx_json:encode(Body)}).
 
 do_request_api(Method, Request)->
     ct:pal("Method: ~p, Request: ~p", [Method, Request]),

changes/v4.3.22-en.md

@@ -15,3 +15,6 @@
 - "Pause due to rate limit" log level demoted from warning to notice [#9134](https://github.com/emqx/emqx/pull/9134).
 
 - Restore old `emqx_auth_jwt` module API, so the hook callback functions registered in older version will not be invalidated after hot-upgrade [#9144](https://github.com/emqx/emqx/pull/9144).
+
+- Fixed the response status code for the `/status` endpoint [#9210](https://github.com/emqx/emqx/pull/9210). 
+  Before the fix, it always returned `200` even if the EMQX application was not running.  Now it returns `503` in that case.

changes/v4.3.22-zh.md

@@ -15,3 +15,6 @@
 - 限速 “Pause due to rate limit” 的日志级别从原先的 `warning` 降级到 `notice` [#9134](https://github.com/emqx/emqx/pull/9134)。
 
 - 保留老的 `emqx_auth_jwt` 模块的接口函数，保障热升级之前添加的回调函数在热升级之后也不会失效 [#9144](https://github.com/emqx/emqx/pull/9144)。
+
+- 修正了 `/status` API 的响应状态代码 [#9210](https://github.com/emqx/emqx/pull/9210)。
+  在修复之前，它总是返回 `200`，即使 EMQX 应用程序没有运行。 现在它在这种情况下返回 `503`。