fix(iotdb): normalized the timeout configs of the Thrift driver

apps/emqx_bridge_iotdb/src/emqx_bridge_iotdb_connector.erl

@@ -75,7 +75,9 @@
 
 -define(CONNECTOR_TYPE, iotdb).
 -define(IOTDB_PING_PATH, <<"ping">>).
--define(DEFAULT_THRIFT_TIMEOUT, timer:seconds(10)).
+
+%% timer:seconds(10)).
+-define(DEFAULT_THRIFT_TIMEOUT, 10000).
 
 -import(hoconsc, [mk/2, enum/1, ref/2]).
 
@@ -336,9 +338,10 @@ on_start(
 
     DriverOpts = maps:merge(
         #{
-            connect_timeout => ?DEFAULT_THRIFT_TIMEOUT, recv_timeout => ?DEFAULT_THRIFT_TIMEOUT
+            connect_timeout => ?DEFAULT_THRIFT_TIMEOUT,
+            recv_timeout => ?DEFAULT_THRIFT_TIMEOUT
         },
-        maps:with([connect_timeout, recv_timeout], Config)
+        normalize_thrift_timeout(maps:with([connect_timeout, recv_timeout], Config))
     ),
 
     DriverOpts1 =
@@ -1103,3 +1106,25 @@ do_on_query(InstanceId, ChannelId, Data, #{driver := restapi} = State) ->
     emqx_bridge_http_connector:on_query(InstanceId, {ChannelId, Data}, State);
 do_on_query(InstanceId, _ChannelId, Data, #{driver := thrift} = _State) ->
     ecpool:pick_and_do(InstanceId, {iotdb, insert_tablet, [Data]}, no_handover).
+
+%% 1. The default timeout in Thrift is `infinity`, but it may cause stuck
+%% 2. The schema of `timeout` accepts a zero value, but the Thrift driver not
+%% 3. If the timeout is too small, the driver may not work properly
+normalize_thrift_timeout(Timeouts) ->
+    maps:map(
+        fun
+            (_K, V) when V >= ?DEFAULT_THRIFT_TIMEOUT ->
+                V;
+            (K, V) ->
+                ?SLOG(warning, #{
+                    msg => "iotdb_thrift_timeout_normalized",
+                    reason => "The timeout is too small for the Thrift driver to work",
+                    timeout => K,
+                    from => V,
+                    to => ?DEFAULT_THRIFT_TIMEOUT,
+                    unit => millisecond
+                }),
+                ?DEFAULT_THRIFT_TIMEOUT
+        end,
+        Timeouts
+    ).