简化service的逻辑

apps/efka/src/efka_inetd.erl

@@ -17,7 +17,7 @@
 
 %% API
 -export([start_link/0]).
--export([deploy/3]).
+-export([deploy/3, start_service/1, stop_service/1]).
 
 %% gen_server callbacks
 -export([init/1, handle_call/3, handle_cast/2, handle_info/2, terminate/2, code_change/3]).
@@ -38,6 +38,14 @@
 deploy(TaskId, ServerId, TarUrl) when is_integer(TaskId), is_binary(ServerId), is_binary(TarUrl) ->
     gen_server:call(?SERVER, {deploy, TaskId, ServerId, TarUrl}).
 
+-spec start_service(ServiceId :: binary()) -> ok | {error, Reason :: term()}.
+start_service(ServiceId) when is_binary(ServiceId) ->
+    gen_server:call(?SERVER, {start_service, ServiceId}).
+
+-spec stop_service(ServiceId :: binary()) -> ok | {error, Reason :: term()}.
+stop_service(ServiceId) when is_binary(ServiceId) ->
+    gen_server:call(?SERVER, {stop_service, ServiceId}).
+
 %% @doc Spawns the server and registers the local name (unique)
 -spec(start_link() ->
     {ok, Pid :: pid()} | ignore | {error, Reason :: term()}).
@@ -73,6 +81,8 @@ handle_call({deploy, TaskId, ServiceId, TarUrl}, _From, State = #state{root_dir
     {ok, ServiceRootDir} = ensure_dirs(RootDir, ServiceId),
 
     ServicePid = efka_service:get_pid(ServiceId),
+    lager:debug("service pid is: ~p", [ServicePid]),
+
     case is_pid(ServicePid) andalso efka_service:is_running(ServicePid) of
         true ->
             {reply, {error, <<"the service is running, stop first">>}, State};
@@ -90,6 +100,34 @@ handle_call({deploy, TaskId, ServiceId, TarUrl}, _From, State = #state{root_dir
             end
     end;
 
+%% 启动服务: 当前服务如果正常运行，则不允许重启
+handle_call({start_service, ServiceId}, _From, State) ->
+    case efka_service:get_pid(ServiceId) of
+        undefined ->
+            case efka_service_sup:start_service(ServiceId) of
+                {ok, _} ->
+                    %% 更新数据库状态, 状态是为了保证下次efka重启的时候，服务能够启动
+                    ok = service_model:change_status(ServiceId, 1),
+                    {reply, ok, State};
+                {error, Reason} ->
+                    {reply, {error, Reason}, State}
+            end;
+        ServicePid when is_pid(ServicePid) ->
+            {reply, {error, <<"service is running">>}, State}
+    end;
+
+%% 停止服务, 主动停止的时候会改变服务配置的status字段
+handle_call({stop_service, ServiceId}, _From, State = #state{}) ->
+    case efka_service:get_pid(ServiceId) of
+        undefined ->
+            {reply, {error, <<"service not running">>}, State};
+        ServicePid ->
+            efka_service_sup:delete_service(ServicePid),
+            %% 主动停止的服务，需要更新数据库状态, 状态是为了保证下次efka重启的时候，不自动启动服务
+            ok = service_model:change_status(ServiceId, 0),
+            {reply, ok, State}
+    end;
+
 handle_call(_Request, _From, State = #state{}) ->
     {reply, ok, State}.
 

apps/efka/src/efka_manifest.erl

@@ -41,7 +41,7 @@ new(ServiceRootDir) when is_list(ServiceRootDir) ->
     end.
 
 -spec startup(Manifest :: #manifest{}) -> {ok, Port :: port()} | {error, Reason :: binary()}.
-startup(#manifest{work_dir = WorkDir, exec = ExecCmd0, args = Args0}) ->
+startup(#manifest{id = Id, work_dir = WorkDir, exec = ExecCmd0, args = Args0}) ->
     PortSettings = [
         {cd, WorkDir},
         {args, [binary_to_list(A) || A <- Args0]},
@@ -49,6 +49,7 @@ startup(#manifest{work_dir = WorkDir, exec = ExecCmd0, args = Args0}) ->
     ],
     ExecCmd = binary_to_list(ExecCmd0),
     RealExecCmd = filename:absname_join(WorkDir, ExecCmd),
+    lager:debug("[efka_manifest] service_id: ~p, real command is: ~p", [Id, RealExecCmd]),
     case catch erlang:open_port({spawn_executable, RealExecCmd}, PortSettings) of
         Port when is_port(Port) ->
             {ok, Port};

apps/efka/src/efka_service.erl

@@ -13,14 +13,10 @@
 
 -behaviour(gen_server).
 
-%% 当前微服务的状态
--define(STATUS_STOPPED, stopped).
--define(STATUS_RUNNING, running).
-
 %% API
 -export([start_link/2]).
 -export([get_name/1, get_pid/1, start_service/1, stop_service/1, attach_channel/2]).
--export([push_config/3, request_config/1, invoke/3, is_running/1]).
+-export([push_config/3, request_config/1, invoke/3]).
 -export([metric_data/3, send_event/3]).
 
 %% gen_server callbacks
@@ -36,9 +32,7 @@
     os_pid :: undefined | integer(),
     %% 配置信息
     manifest :: undefined | efka_manifest:manifest(),
-    inflight = #{},
+    inflight = #{}
-    %% 当前服务的运行状态
-    running_status = ?STATUS_STOPPED
 }).
 
 %%%===================================================================
@@ -59,10 +53,6 @@ push_config(Pid, Ref, ConfigJson) when is_pid(Pid), is_binary(ConfigJson) ->
 invoke(Pid, Ref, Payload) when is_pid(Pid), is_reference(Ref), is_binary(Payload) ->
     gen_server:cast(Pid, {invoke, Ref, self(), Payload}).
 
--spec is_running(Pid :: pid()) -> boolean().
-is_running(Pid) when is_pid(Pid) ->
-    gen_server:call(Pid, is_running).
-
 request_config(Pid) when is_pid(Pid) ->
     gen_server:call(Pid, request_config).
 
@@ -104,31 +94,25 @@ init([ServiceId]) ->
         error ->
             lager:notice("[efka_service] service_id: ~p, not found", [ServiceId]),
             ignore;
-        {ok, Service = #service{root_dir = RootDir}} ->
+        {ok, #service{root_dir = RootDir}} ->
+            %% 第一次启动，要求必须成功；只有第一次启动成功，后续的重启逻辑才有意义
             case efka_manifest:new(RootDir) of
                 {ok, Manifest} ->
-                    init0(Service, Manifest);
+                    case efka_manifest:startup(Manifest) of
+                        {ok, Port} ->
+                            {os_pid, OSPid} = erlang:port_info(Port, os_pid),
+                            lager:debug("[efka_service] service: ~p, port: ~p, boot_service success os_pid: ~p", [ServiceId, Port, OSPid]),
+                            {ok, #state{service_id = ServiceId, manifest = Manifest, port = Port, os_pid = OSPid}};
+                        {error, Reason} ->
+                            lager:debug("[efka_service] service: ~p, boot_service get error: ~p", [ServiceId, Reason]),
+                            {stop, Reason}
+                    end;
                 {error, Reason} ->
                     lager:notice("[efka_service] service: ~p, read manifest.json get error: ~p", [ServiceId, Reason]),
                     ignore
             end
     end.
 
-init0(#service{service_id = ServiceId, status = 1}, Manifest) ->
-    %% 数据的状态和运行状态是2回事
-    case efka_manifest:startup(Manifest) of
-        {ok, Port} ->
-            {os_pid, OSPid} = erlang:port_info(Port, os_pid),
-            lager:debug("[efka_service] service: ~p, port: ~p, boot_service success os_pid: ~p", [ServiceId, Port, OSPid]),
-            {ok, #state{service_id = ServiceId, manifest = Manifest, running_status = ?STATUS_RUNNING, port = Port, os_pid = OSPid}};
-        {error, Reason} ->
-            lager:debug("[efka_service] service: ~p, boot_service get error: ~p", [ServiceId, Reason]),
-            {ok, #state{service_id = ServiceId, manifest = Manifest, running_status = ?STATUS_STOPPED, port = undefined, os_pid = undefined}}
-    end;
-init0(#service{service_id = ServiceId, status = 0}, Manifest) ->
-    lager:debug("[efka_service] service: ~p current status is 0, not boot", [ServiceId]),
-    {ok, #state{service_id = ServiceId, manifest = Manifest, running_status = ?STATUS_STOPPED, port = undefined, os_pid = undefined}}.
-
 %% @private
 %% @doc Handling call messages
 -spec(handle_call(Request :: term(), From :: {pid(), Tag :: term()},
@@ -140,59 +124,20 @@ init0(#service{service_id = ServiceId, status = 0}, Manifest) ->
     {stop, Reason :: term(), Reply :: term(), NewState :: #state{}} |
     {stop, Reason :: term(), NewState :: #state{}}).
 %% 绑定channel
-handle_call({attach_channel, ChannelPid}, _From, State = #state{channel_pid = OldChannelPid, service_id = ServiceId}) ->
+handle_call({attach_channel, ChannelPid}, _From, State = #state{channel_pid = OldChannelPid}) ->
-    Status = service_model:get_status(ServiceId),
+    case is_pid(OldChannelPid) andalso is_process_alive(OldChannelPid) of
-    case {Status, is_pid(OldChannelPid) andalso is_process_alive(OldChannelPid)} of
+        false ->
-        {1, false} ->
             erlang:monitor(process, ChannelPid),
             {reply, ok, State#state{channel_pid = ChannelPid}};
-        {1, true} ->
+        true ->
-            {reply, {error, <<"channel exists">>}, State};
+            {reply, {error, <<"channel exists">>}, State}
-        {0, _} ->
-            {reply, {error, <<"serivce stopped">>}, State}
     end;
 
-%% 获取服务的运行状态
-handle_call(is_running, _From, State = #state{running_status = RunningStatus}) ->
-    {reply, RunningStatus, State};
-
 %% 请求参数项 done
-handle_call(request_config, _From, State = #state{service_id = ServiceId, running_status = ?STATUS_RUNNING}) ->
+handle_call(request_config, _From, State = #state{service_id = ServiceId}) ->
     Params = service_model:get_params(ServiceId),
     {reply, {ok, Params}, State};
 
-%% 启动服务: 当前服务如果正常运行，则不允许重启
-handle_call(start_service, _From, State = #state{running_status = ?STATUS_RUNNING}) ->
-    {reply, {error, <<"service is running">>}, State};
-handle_call(start_service, _From, State = #state{running_status = ?STATUS_STOPPED, manifest = Manifest, service_id = ServiceId}) ->
-    %% 异步启动服务
-    case efka_manifest:startup(Manifest) of
-        {ok, Port} ->
-            {os_pid, OSPid} = erlang:port_info(Port, os_pid),
-            lager:debug("[efka_service] service_id: ~p, start_service port: ~p, os_pid: ~p", [ServiceId, Port, OSPid]),
-            %% 更新数据库状态
-            ok = service_model:change_status(ServiceId, 1),
-            {reply, ok, State#state{running_status = ?STATUS_RUNNING, port = Port, os_pid = OSPid}};
-        {error, Reason} ->
-            %% 启动失败不能更新数据库里面的状态
-            {reply, {error, Reason}, State}
-    end;
-
-%% 停止服务, 主动停止的时候会改变服务配置的status字段
-handle_call(stop_service, _From, State = #state{running_status = ?STATUS_STOPPED, service_id = ServiceId, port = Port, os_pid = OSPid}) ->
-    lager:debug("[efka_service] service_id: ~p, stop service port: ~p, os_pid: ~p", [ServiceId, Port, OSPid]),
-    {reply, {error, <<"service not running">>}, State};
-
-handle_call(stop_service, _From, State = #state{running_status = ?STATUS_RUNNING, port = Port, os_pid = OSPid, service_id = ServiceId}) when is_port(Port) ->
-    %% 优先使用微服务提供的stop指令, 没有提供的情况下，使用kill指令
-    kill_os_pid(OSPid),
-
-    erlang:is_port(Port) andalso erlang:port_close(Port),
-    lager:debug("[efka_service] service_id: ~p, port: ~p, os_pid: ~p, will closed", [ServiceId, Port, OSPid]),
-    ok = service_model:change_status(ServiceId, 0),
-
-    {reply, ok, State#state{port = undefined, os_pid = undefined, running_status = ?STATUS_STOPPED}};
-
 handle_call(_Request, _From, State = #state{}) ->
     {reply, ok, State}.
 
@@ -211,7 +156,7 @@ handle_cast({send_event, EventType, Params}, State = #state{service_id = Service
     {noreply, State};
 
 %% 推送配置项目
-handle_cast({push_config, Ref, ReceiverPid, ConfigJson}, State = #state{running_status = ?STATUS_RUNNING, channel_pid = ChannelPid, inflight = Inflight}) ->
+handle_cast({push_config, Ref, ReceiverPid, ConfigJson}, State = #state{channel_pid = ChannelPid, inflight = Inflight}) ->
     case is_pid(ChannelPid) andalso is_process_alive(ChannelPid) of
         true ->
             efka_tcp_channel:push_config(ChannelPid, Ref, self(), ConfigJson),
@@ -222,7 +167,7 @@ handle_cast({push_config, Ref, ReceiverPid, ConfigJson}, State = #state{running_
     end;
 
 %% 推送配置项目
-handle_cast({invoke, Ref, ReceiverPid, Payload}, State = #state{running_status = ?STATUS_RUNNING, channel_pid = ChannelPid, inflight = Inflight}) ->
+handle_cast({invoke, Ref, ReceiverPid, Payload}, State = #state{channel_pid = ChannelPid, inflight = Inflight}) ->
     case is_pid(ChannelPid) andalso is_process_alive(ChannelPid) of
         true ->
             efka_tcp_channel:invoke(ChannelPid, Ref, self(), Payload),
@@ -243,20 +188,15 @@ handle_cast(_Request, State = #state{}) ->
     {stop, Reason :: term(), NewState :: #state{}}).
 %% 重启服务
 handle_info({timeout, _, reboot_service}, State = #state{service_id = ServiceId, manifest = Manifest}) ->
-    case service_model:get_status(ServiceId) of
+    case efka_manifest:startup(Manifest) of
-        0 ->
+        {ok, Port} ->
-            lager:debug("[efka_service] service_id: ~p, is stopped, ignore boot_service", [ServiceId]),
+            {os_pid, OSPid} = erlang:port_info(Port, os_pid),
-            {noreply, State};
+            lager:debug("[efka_service] service_id: ~p, reboot success: ~p, port: ~p, os_pid: ~p", [ServiceId, Port, OSPid]),
-        1 ->
+            {noreply, State#state{port = Port, os_pid = OSPid}};
-            case efka_manifest:startup(Manifest) of
+        {error, Reason} ->
-                {ok, Port} ->
+            lager:debug("[efka_service] service_id: ~p, boot_service get error: ~p", [ServiceId, Reason]),
-                    {os_pid, OSPid} = erlang:port_info(Port, os_pid),
+            try_reboot(),
-                    lager:debug("[efka_service] service_id: ~p, reboot success: ~p, port: ~p, os_pid: ~p", [ServiceId, Port, OSPid]),
+            {noreply, State}
-                    {noreply, State#state{running_status = ?STATUS_RUNNING, port = Port, os_pid = OSPid}};
-                {error, Reason} ->
-                    lager:debug("[efka_service] service_id: ~p, boot_service get error: ~p", [ServiceId, Reason]),
-                    {noreply, State#state{running_status = ?STATUS_STOPPED}}
-            end
     end;
 
 %% 处理channel的回复
@@ -276,8 +216,8 @@ handle_info({Port, {data, Data}}, State = #state{port = Port, service_id = Servi
 %% 处理port的消息, Port的被动关闭会触发；因此这个时候的Port和State.port的值是相等的
 handle_info({Port, {exit_status, Code}}, State = #state{service_id = ServiceId}) ->
     lager:debug("[efka_service] service_id: ~p, port: ~p, exit with code: ~p", [ServiceId, Port, Code]),
-    % erlang:start_timer(5000, self(), reboot_service),
+    try_reboot(),
-    {noreply, State#state{port = undefined, os_pid = undefined, running_status = ?STATUS_STOPPED}};
+    {noreply, State#state{port = undefined, os_pid = undefined}};
 
 %% 处理channel进程的退出
 handle_info({'DOWN', _Ref, process, ChannelPid, Reason}, State = #state{channel_pid = ChannelPid, service_id = ServiceId}) ->
@@ -291,11 +231,9 @@ handle_info({'DOWN', _Ref, process, ChannelPid, Reason}, State = #state{channel_
 %% with Reason. The return value is ignored.
 -spec(terminate(Reason :: (normal | shutdown | {shutdown, term()} | term()),
     State :: #state{}) -> term()).
-terminate(Reason, _State = #state{service_id = ServiceId, os_pid = OSPid}) ->
+terminate(Reason, _State = #state{service_id = ServiceId, port = Port, os_pid = OSPid}) ->
-    lager:debug("[efka_service] service_id: ~p, terminate with reason: ~p", [ServiceId, Reason]),
+    erlang:is_port(Port) andalso erlang:port_close(Port),
     kill_os_pid(OSPid),
-    ok;
-terminate(Reason, #state{service_id = ServiceId}) ->
     lager:debug("[efka_service] service_id: ~p, terminate with reason: ~p", [ServiceId, Reason]),
     ok.
 
@@ -318,4 +256,8 @@ kill_os_pid(undefined) ->
 kill_os_pid(OSPid) when is_integer(OSPid) ->
     Cmd = lists:flatten(io_lib:format("kill -9 ~p", [OSPid])),
     lager:debug("kill cmd is: ~p", [Cmd]),
-    os:cmd(Cmd).
+    os:cmd(Cmd).
+
+-spec try_reboot() -> no_return().
+try_reboot() ->
+    erlang:start_timer(5000, self(), reboot_service).

apps/efka/src/efka_service_sup.erl

@@ -46,8 +46,9 @@ start_link() ->
     | ignore | {error, Reason :: term()}).
 init([]) ->
     SupFlags = #{strategy => one_for_one, intensity => 1000, period => 3600},
-    MicroServiceIds = service_model:get_all_service_ids(),
+    %% 简化逻辑，只启动需要运行的微服务
-    Specs = lists:map(fun(ServiceId) -> child_spec(ServiceId) end, MicroServiceIds),
+    Services = service_model:get_running_services(),
+    Specs = lists:map(fun(ServiceId) -> child_spec(ServiceId) end, Services),
 
     {ok, {SupFlags, Specs}}.
 
@@ -72,13 +73,15 @@ delete_service(ServiceId) when is_binary(ServiceId) ->
     ok = supervisor:terminate_child(?MODULE, ChildId),
     supervisor:delete_child(?MODULE, ChildId).
 
+child_spec(#service{service_id = ServiceId}) when is_binary(ServiceId) ->
+    child_spec(ServiceId);
 child_spec(ServiceId) when is_binary(ServiceId) ->
     Name = efka_service:get_name(ServiceId),
     #{
         id => Name,
         start => {efka_service, start_link, [Name, ServiceId]},
         restart => permanent,
-        shutdown => 2000,
+        shutdown => 5000,
         type => worker,
         modules => ['efka_service']
     }.

apps/efka/src/mnesia/service_model.erl

@@ -15,7 +15,7 @@
 
 %% API
 -export([create_table/0]).
--export([insert/1, get_all_services/0, get_all_service_ids/0]).
+-export([insert/1, get_all_services/0, get_all_service_ids/0, get_running_services/0]).
 -export([get_metrics/1, get_params/1, set_metrics/2, set_params/2, get_service/1, get_status/1, change_status/2]).
 
 create_table() ->
@@ -137,4 +137,17 @@ get_all_services() ->
 
 -spec get_all_service_ids() -> [ServiceId :: binary()].
 get_all_service_ids() ->
-    mnesia:dirty_all_keys(?TAB).
+    mnesia:dirty_all_keys(?TAB).
+
+-spec get_running_services() -> {ok, [#service{}]} | {error, Reason :: term()}.
+get_running_services() ->
+    F = fun() ->
+        Q = qlc:q([E || E <- mnesia:table(?TAB), E#service.status == 1]),
+        qlc:e(Q)
+        end,
+    case mnesia:transaction(F) of
+        {atomic, Services} ->
+            {ok, Services};
+        {aborted, Error} ->
+            {error, Error}
+    end.