roll back to (almost) default - keeping the extra error information when

corrupt cluster status files are found

2 files changed, 9 insertions, 32 deletions

diff --git a/src/rabbit.erl b/src/rabbit.erl
index f3d31b2244..ef9f5f56ce 100644
--- a/src/rabbit.erl
+++ b/src/rabbit.erl
@@ -174,16 +174,10 @@
                    [{mfa,         {rabbit_networking, boot, []}},
                     {requires,    log_relay}]}).
 
--rabbit_boot_step({app_running,
-                   [{description, "cluster membership"},
-                    {mfa,         {rabbit_sup, start_restartable_child,
-                                   [rabbit_app_marker]}},
-                    {requires,    networking}]}).
-
 -rabbit_boot_step({notify_cluster,
                    [{description, "notify cluster nodes"},
                     {mfa,         {rabbit_node_monitor, notify_node_up, []}},
-                    {requires,    app_running}]}).
+                    {requires,    networking}]}).
 
 %%---------------------------------------------------------------------------
 
diff --git a/src/rabbit_node_monitor.erl b/src/rabbit_node_monitor.erl
index b89d4dc4bc..97feb2f2b4 100644
--- a/src/rabbit_node_monitor.erl
+++ b/src/rabbit_node_monitor.erl
@@ -112,7 +112,7 @@ prepare_cluster_status_files() ->
                     true  -> ThisNode;
                     false -> []
                 end,
-    write_cluster_status({AllNodes2, DiscNodes, RunningNodes2}).
+    ok = write_cluster_status({AllNodes2, DiscNodes, RunningNodes2}).
 
 write_cluster_status({All, Disc, Running}) ->
     ClusterStatusFN = cluster_status_filename(),
@@ -199,7 +199,7 @@ handle_call(_Request, _From, State) ->
 %% mnesia propagation.
 handle_cast({node_up, Node, NodeType},
             State = #state{monitors = Monitors}) ->
-    case pmon:is_monitored({rabbit_running, Node}, Monitors) of
+    case pmon:is_monitored({rabbit, Node}, Monitors) of
         true  -> {noreply, State};
         false -> rabbit_log:info("rabbit on node ~p up~n", [Node]),
                  {AllNodes, DiscNodes, RunningNodes} = read_cluster_status(),
@@ -210,12 +210,8 @@ handle_cast({node_up, Node, NodeType},
                                        end,
                                        add_node(Node, RunningNodes)}),
                  ok = handle_live_rabbit(Node),
-                 State1 = mon({rabbit_running, Node}, State),
-                 State2 = case pmon:is_monitored({rabbit, Node}, Monitors) of
-                              true  -> State1;
-                              false -> mon({rabbit, Node}, State1)
-                          end,
-                 {noreply, State2}
+                 {noreply, State#state{
+                             monitors = pmon:monitor({rabbit, Node}, Monitors)}}
     end;
 handle_cast({joined_cluster, Node, NodeType}, State) ->
     {AllNodes, DiscNodes, RunningNodes} = read_cluster_status(),
@@ -234,20 +230,13 @@ handle_cast({left_cluster, Node}, State) ->
 handle_cast(_Msg, State) ->
     {noreply, State}.
 
-handle_info({'DOWN', _MRef, process, {rabbit_running, Node}, _Reason}, State) ->
-    %% The node has started to stop, remove it from the cluster status
-    %% file. We want to do this "early" to stand a better chance of
-    %% recording anything when all the nodes are shut down
-    %% simultaneously.
+handle_info({'DOWN', _MRef, process, {rabbit, Node}, _Reason},
+            State = #state{monitors = Monitors}) ->
+    rabbit_log:info("rabbit on node ~p down~n", [Node]),
     {AllNodes, DiscNodes, RunningNodes} = read_cluster_status(),
     write_cluster_status({AllNodes, DiscNodes, del_node(Node, RunningNodes)}),
-    {noreply, unmon({rabbit_running, Node}, State)};
-
-handle_info({'DOWN', _MRef, process, {rabbit, Node}, _Reason}, State) ->
-    %% The node has finished stopping (rabbit anyway), treat it as dead.
-    rabbit_log:info("rabbit on node ~p down~n", [Node]),
     ok = handle_dead_rabbit(Node),
-    {noreply, unmon({rabbit, Node}, State)};
+    {noreply, State#state{monitors = pmon:erase({rabbit, Node}, Monitors)}};
 
 handle_info({mnesia_system_event,
              {inconsistent_database, running_partitioned_network, Node}},
@@ -305,9 +294,3 @@ legacy_should_be_disc_node(DiscNodes) ->
 add_node(Node, Nodes) -> lists:usort([Node | Nodes]).
 
 del_node(Node, Nodes) -> Nodes -- [Node].
-
-mon(Item, State = #state{monitors = Monitors}) ->
-    State#state{monitors = pmon:monitor(Item, Monitors)}.
-
-unmon(Item, State = #state{monitors = Monitors}) ->
-    State#state{monitors = pmon:erase(Item, Monitors)}.