diff --git a/apps/emqx_resource/src/emqx_resource_buffer_worker.erl b/apps/emqx_resource/src/emqx_resource_buffer_worker.erl
index 2e2cd5631..7cb7f8198 100644
--- a/apps/emqx_resource/src/emqx_resource_buffer_worker.erl
+++ b/apps/emqx_resource/src/emqx_resource_buffer_worker.erl
@@ -1466,7 +1466,7 @@ mark_inflight_items_as_retriable(Data, WorkerMRef) ->
             end
         ),
     _NumAffected = ets:select_replace(InflightTID, MatchSpec),
-    ?tp(buffer_worker_worker_down_update, #{num_affected => _NumAffected}),
+    ?tp(buffer_worker_async_agent_down, #{num_affected => _NumAffected}),
     ok.
 
 %% used to update a batch after dropping expired individual queries.
diff --git a/apps/emqx_resource/test/emqx_resource_SUITE.erl b/apps/emqx_resource/test/emqx_resource_SUITE.erl
index f8ddd56b5..34781df6c 100644
--- a/apps/emqx_resource/test/emqx_resource_SUITE.erl
+++ b/apps/emqx_resource/test/emqx_resource_SUITE.erl
@@ -1766,12 +1766,6 @@ t_async_pool_worker_death(_Config) ->
             ?assertEqual(NumReqs, Inflight0),
 
             %% grab one of the worker pids and kill it
-            {ok, SRef1} =
-                snabbkaffe:subscribe(
-                    ?match_event(#{?snk_kind := buffer_worker_worker_down_update}),
-                    NumBufferWorkers,
-                    10_000
-                ),
             {ok, #{pid := Pid0}} = emqx_resource:simple_sync_query(?ID, get_state),
             MRef = monitor(process, Pid0),
             ct:pal("will kill ~p", [Pid0]),
@@ -1785,13 +1779,27 @@ t_async_pool_worker_death(_Config) ->
             end,
 
             %% inflight requests should have been marked as retriable
-            {ok, _} = snabbkaffe:receive_events(SRef1),
+            wait_until_all_marked_as_retriable(NumReqs),
             Inflight1 = emqx_resource_metrics:inflight_get(?ID),
             ?assertEqual(NumReqs, Inflight1),
 
-            ok
+            NumReqs
         end,
-        []
+        fun(NumReqs, Trace) ->
+            Events = ?of_kind(buffer_worker_async_agent_down, Trace),
+            %% At least one buffer worker should have marked its
+            %% requests as retriable.  If a single one has
+            %% received all requests, that's all we got.
+            ?assertMatch([_ | _], Events),
+            %% All requests distributed over all buffer workers
+            %% should have been marked as retriable, by the time
+            %% the inflight has been drained.
+            ?assertEqual(
+                NumReqs,
+                lists:sum([N || #{num_affected := N} <- Events])
+            ),
+            ok
+        end
     ),
     ok.
 
@@ -3017,3 +3025,33 @@ trace_between_span(Trace0, Marker) ->
     {Trace1, [_ | _]} = ?split_trace_at(#{?snk_kind := Marker, ?snk_span := {complete, _}}, Trace0),
     {[_ | _], [_ | Trace2]} = ?split_trace_at(#{?snk_kind := Marker, ?snk_span := start}, Trace1),
     Trace2.
+
+wait_until_all_marked_as_retriable(NumExpected) when NumExpected =< 0 ->
+    ok;
+wait_until_all_marked_as_retriable(NumExpected) ->
+    Seen = #{},
+    do_wait_until_all_marked_as_retriable(NumExpected, Seen).
+
+do_wait_until_all_marked_as_retriable(NumExpected, _Seen) when NumExpected =< 0 ->
+    ok;
+do_wait_until_all_marked_as_retriable(NumExpected, Seen) ->
+    Res = ?block_until(
+        #{?snk_kind := buffer_worker_async_agent_down, ?snk_meta := #{pid := P}} when
+            not is_map_key(P, Seen),
+        10_000
+    ),
+    case Res of
+        {timeout, Evts} ->
+            ct:pal("events so far:\n  ~p", [Evts]),
+            ct:fail("timeout waiting for events");
+        {ok, #{num_affected := NumAffected, ?snk_meta := #{pid := Pid}}} ->
+            ct:pal("affected: ~p; pid: ~p", [NumAffected, Pid]),
+            case NumAffected >= NumExpected of
+                true ->
+                    ok;
+                false ->
+                    do_wait_until_all_marked_as_retriable(NumExpected - NumAffected, Seen#{
+                        Pid => true
+                    })
+            end
+    end.