Don't perform CMQ leadership transfer when entering maintenance mode

The time this operation can take in clusters with a lot of classic
mirrored queue (say, 10s or 100s of thousands) be prohibitive for
upgrades.

Upgrades that use a health check to ensure that there are in-sync
replicas before entering maintenance mode, in which case
the transfer is not really necessary.

All of the above is more obvious with the recent changes in #2749.

deps/rabbit/src/rabbit_maintenance.erl

@@ -88,7 +88,8 @@ do_drain() ->
     ReadableCandidates = readable_candidate_list(TransferCandidates),
     rabbit_log:info("Node will transfer primary replicas of its queues to ~b peers: ~s",
                     [length(TransferCandidates), ReadableCandidates]),
-    transfer_leadership_of_classic_mirrored_queues(TransferCandidates),
+    %% Note: only QQ leadership is transferred because it is a reasonably quick thing to do a lot of queues
+    %% in the cluster, unlike with CMQs.
     transfer_leadership_of_quorum_queues(TransferCandidates),
     stop_local_quorum_queue_followers(),
 
@@ -248,7 +249,12 @@ transfer_leadership_of_quorum_queues(_TransferCandidates) ->
     rabbit_log:info("Leadership transfer for quorum queues hosted on this node has been initiated").
 
 -spec transfer_leadership_of_classic_mirrored_queues([node()]) -> ok.
- transfer_leadership_of_classic_mirrored_queues([]) ->
+%% This function is no longer used by maintanence mode. We retain it in case
+%% classic mirrored queue leadership transfer would be reconsidered.
+%%
+%% With a lot of CMQs in a cluster, the transfer procedure can take prohibitively long
+%% for a pre-upgrade task.
+transfer_leadership_of_classic_mirrored_queues([]) ->
     rabbit_log:warning("Skipping leadership transfer of classic mirrored queues: no candidate "
                        "(online, not under maintenance) nodes to transfer to!");
 transfer_leadership_of_classic_mirrored_queues(TransferCandidates) ->

deps/rabbit/test/maintenance_mode_SUITE.erl

@@ -24,8 +24,7 @@ groups() ->
       {cluster_size_3, [], [
           maintenance_mode_status,
           listener_suspension_status,
-          client_connection_closure,
+          client_connection_closure
-          classic_mirrored_queue_leadership_transfer
         ]},
       {quorum_queues, [], [
           quorum_queue_leadership_transfer
@@ -211,33 +210,6 @@ client_connection_closure(Config) ->
     rabbit_ct_broker_helpers:revive_node(Config, A).
 
 
-classic_mirrored_queue_leadership_transfer(Config) ->
-    [A | _] = rabbit_ct_broker_helpers:get_node_configs(Config, nodename),
-    ct:pal("Picked node ~s for maintenance tests...", [A]),
-
-    rabbit_ct_helpers:await_condition(
-        fun () -> not rabbit_ct_broker_helpers:is_being_drained_local_read(Config, A) end, 10000),
-
-    PolicyPattern = <<"^cq.mirrored">>,
-    rabbit_ct_broker_helpers:set_ha_policy(Config, A, PolicyPattern, <<"all">>),
-
-    Conn = rabbit_ct_client_helpers:open_connection(Config, A),
-    {ok, Ch} = amqp_connection:open_channel(Conn),
-    QName = <<"cq.mirrored.1">>,
-    amqp_channel:call(Ch, #'queue.declare'{queue = QName, durable = true}),
-
-    ?assertEqual(1, length(rabbit_ct_broker_helpers:rpc(Config, A, rabbit_amqqueue, list_local, [<<"/">>]))),
-
-    rabbit_ct_broker_helpers:drain_node(Config, A),
-    rabbit_ct_helpers:await_condition(
-        fun () -> rabbit_ct_broker_helpers:is_being_drained_local_read(Config, A) end, 10000),
-
-    ?assertEqual(0, length(rabbit_ct_broker_helpers:rpc(Config, A, rabbit_amqqueue, list_local, [<<"/">>]))),
-
-    rabbit_ct_broker_helpers:revive_node(Config, A),
-    %% rabbit_ct_broker_helpers:set_ha_policy/4 uses pattern for policy name
-    rabbit_ct_broker_helpers:clear_policy(Config, A, PolicyPattern).
-
 quorum_queue_leadership_transfer(Config) ->
     [A | _] = Nodenames = rabbit_ct_broker_helpers:get_node_configs(
                             Config, nodename),