Merge pull request #9797 from rabbitmq/use-join_cluster-in-peer-discovery

rabbit_peer_discovery: Rewrite the core logic

Author: dumbbell

deps/rabbit/BUILD.bazel

@@ -958,6 +958,14 @@ rabbitmq_suite(
     ],
 )
 
+rabbitmq_suite(
+    name = "unit_cluster_formation_sort_nodes_SUITE",
+    size = "small",
+    deps = [
+        "@meck//:erlang_app",
+    ],
+)
+
 rabbitmq_suite(
     name = "unit_collections_SUITE",
     size = "small",

deps/rabbit/app.bzl

@@ -1735,6 +1735,14 @@ def test_suite_beam_files(name = "test_suite_beam_files"):
         app_name = "rabbit",
         erlc_opts = "//:test_erlc_opts",
     )
+    erlang_bytecode(
+        name = "unit_cluster_formation_sort_nodes_SUITE_beam_files",
+        testonly = True,
+        srcs = ["test/unit_cluster_formation_sort_nodes_SUITE.erl"],
+        outs = ["test/unit_cluster_formation_sort_nodes_SUITE.beam"],
+        app_name = "rabbit",
+        erlc_opts = "//:test_erlc_opts",
+    )
     erlang_bytecode(
         name = "unit_collections_SUITE_beam_files",
         testonly = True,

deps/rabbit/src/rabbit_db.erl

@@ -51,11 +51,22 @@ init() ->
        #{domain => ?RMQLOG_DOMAIN_DB}),
 
     ensure_dir_exists(),
-    rabbit_peer_discovery:log_configured_backend(),
     rabbit_peer_discovery:maybe_init(),
 
     pre_init(IsVirgin),
 
+    case IsVirgin of
+        true ->
+            %% At this point, the database backend could change if the node
+            %% joins a cluster and that cluster uses a different database.
+            ?LOG_INFO(
+               "DB: virgin node -> run peer discovery",
+               #{domain => ?RMQLOG_DOMAIN_DB}),
+            rabbit_peer_discovery:sync_desired_cluster();
+        false ->
+            ok
+    end,
+
     Ret = case rabbit_khepri:is_enabled() of
               true  -> init_using_khepri();
               false -> init_using_mnesia()
@@ -66,8 +77,8 @@ init() ->
                "DB: initialization successeful",
                #{domain => ?RMQLOG_DOMAIN_DB}),
 
+            rabbit_peer_discovery:maybe_register(),
             init_finished(),
-            post_init(IsVirgin),
 
             ok;
         Error ->
@@ -82,12 +93,6 @@ pre_init(IsVirgin) ->
     OtherMembers = rabbit_nodes:nodes_excl_me(Members),
     rabbit_db_cluster:ensure_feature_flags_are_in_sync(OtherMembers, IsVirgin).
 
-post_init(false = _IsVirgin) ->
-    rabbit_peer_discovery:maybe_register();
-post_init(true = _IsVirgin) ->
-    %% Registration handled by rabbit_peer_discovery.
-    ok.
-
 init_using_mnesia() ->
     ?LOG_DEBUG(
       "DB: initialize Mnesia",

deps/rabbit/src/rabbit_mnesia.erl

@@ -114,13 +114,13 @@
 init() ->
     ensure_mnesia_running(),
     ensure_mnesia_dir(),
+    %% Peer discovery may have been a no-op if it decided that all other nodes
+    %% should join this one. Therefore, we need to look at if this node is
+    %% still virgin and finish our init of Mnesia accordingly. In particular,
+    %% this second part crates all our Mnesia tables.
     case is_virgin_node() of
-        true  ->
-            rabbit_log:info("Node database directory at ~ts is empty. "
-                            "Assuming we need to join an existing cluster or initialise from scratch...",
-                            [dir()]),
-            rabbit_peer_discovery:maybe_create_cluster(
-              fun create_cluster_callback/2);
+        true ->
+            init_db_and_upgrade([node()], disc, true, _Retry = true);
         false ->
             NodeType = node_type(),
             case is_node_type_permitted(NodeType) of
@@ -141,23 +141,6 @@ init() ->
     ok = rabbit_node_monitor:global_sync(),
     ok.
 
-create_cluster_callback(none, NodeType) ->
-    DiscNodes = [node()],
-    NodeType1 = case is_node_type_permitted(NodeType) of
-                    false -> disc;
-                    true  -> NodeType
-                end,
-    init_db_and_upgrade(DiscNodes, NodeType1, true, _Retry = true),
-    ok;
-create_cluster_callback(RemoteNode, NodeType) ->
-    {ok, {_, DiscNodes, _}} = discover_cluster0(RemoteNode),
-    NodeType1 = case is_node_type_permitted(NodeType) of
-                    false -> disc;
-                    true  -> NodeType
-                end,
-    init_db_and_upgrade(DiscNodes, NodeType1, true, _Retry = true),
-    ok.
-
 %% Make the node join a cluster. The node will be reset automatically
 %% before we actually cluster it. The nodes provided will be used to
 %% find out about the nodes in the cluster.